Technology Vision 2025

AI: A Declaration
of Autonomy
Is trust the limit of
AI’s limitless possibilities?
Technology Vision 2025 | AI: A Declaration of Autonomy Foreword 2

AI: A Declaration Welcome to our Technology Vision for 2025. This

25th edition of our annual technology trends report
enabler to acting autonomously on behalf of
people--equipping them with the capability to
We believe we can. We see this new age of
technology as an opportunity to inject trust in
of Autonomy arrives at a watershed moment for technology and perform new tasks and perform others better than AI in a systematic manner so that businesses
humanity. As more and more leaders embrace the ever. Consider the possibilities and opportunities to and people can realize its incredible reinvention
Is trust the limit of AI's need to continuously reinvent using tech, data reinvent as AI finds its way into new and unfamiliar potential. Together, we can prepare now for
limitless possibilities? and AI, they, now more than ever, need a deep territories. To truly understand and take advantage a bold future when AI is autonomous and helps
understanding of AI. Why? Because the rate of AI’s of this potential, enterprises will be creating their us achieve more together.
technology diffusion is unprecedented and the own, unique AI cognitive digital brains that will
pace is only increasing—creating new opportunities completely reshape the role technology plays
for reinvention across the enterprise—including across their enterprise and with their people. This
new ways of achieving efficiencies, operating will dramatically upend how enterprise tech systems
the core of businesses, new business models are designed, used and operated; act as a brand
and new ways of engaging with customers. ambassador; and inhabit in the physical world by Julie Sweet Karthik Narain
powering robotic bodies. And when AI is spread Chair and CEO Group Chief Executive
of Technology and CTO
We view AI as the new digital because, like digital, across an organization, it enables people and AI to
it is both a technology and a new way of working. bring out the best in each other.
We believe it will be used in every part of the
enterprise and it will have a network effect on Leaders are aware of the challenges to creating
everything and everyone involved. Its impact this future, which include high up-front
is already real, and as companies continue to investments in their core technologies, data
scale AI—and use generative AI as a catalyst for centricity and quality, and talent and new skills.
reinvention—it will solve new problems, create And chief among these challenges is trust.
new inventions, change how we work and live,
and transform industries and governments. Our research finds that 77% of executives believe
unlocking the true benefits of AI will only be
Accenture research shows that only 36% of possible when it’s built on a foundation of trust.
executives say their organizations have scaled Leaders must build trust in digital systems and
gen AI solutions, and just 13% report achieving the AI models, with customers and the workforce
significant enterprise-level impact. We are by ensuring accuracy, predictability, consistency
actively equipping them to do it faster and more and traceability over and above the responsible
safely as we see 2025 as the year of scaled AI. use of AI. People’s trust in AI that it will perform as
expected and justly—beyond any technical aspect—
This year’s Technology Vision explores a future is an essential component that we must get right.
when AI transitions from being an automation
Technology Vision 2025 | AI: A Declaration of Autonomy Contents 3

Contents

Introduction
AI: A Declaration 01 The Binary Big Bang 02 Your Face, in the Future 03 W
 hen LLMs get 04 The New Learning Loop
of Autonomy their Bodies
Is trust the limit of AI's When AI expands exponentially, Differentiating when every interface How foundation models How people and AI are defining
limitless possibilities? systems are upended looks the same reinvent robotics a virtuous cycle of learning, leading,
and creating

Page 04-08 Page 09-21 Page 22-33 Page 34-46 Page 47-58
Technology Vision 2025 | AI: A Declaration of Autonomy Introduction 4

AI: A Declaration on Artificial General Intelligence (AGI).2,3 And like

of Autonomy before, the race has captivated business leaders,

governments, and the world at large.
Is trust the limit of AI's
But it’s a red herring—a distraction most business
limitless possibilities? leaders can’t afford. Someday, AGI will be hugely
consequential, but today it’s still far away with deep
technical and ethical challenges to address. Instead,
it’s vitally important that leaders see the far more
We are entering a new chapter pressing matter already here: the generalization of
in technology—one shaped by artificial intelligence, which will bring a new level of
a generalization of AI. Today’s autonomy and capability to enterprises’ systems,
workforces, and operations long before AGI comes
proliferation of accessible and into play.
ever-present AI will drive new
levels of autonomy all throughout The Generalization of AI

the business, evolving the ability To understand this generalization of AI, one simply
to reinvent with tech, data and needs to look around and see how rooted AI is
AI. It will bring nearly limitless becoming in our lives. It’s been nearly 30 years
since Kasparov’s game, and now models that could
possibilities for innovation make Deep Blue look like an average player are
and growth, but also challenge sitting in everyone’s pockets. The Turing Test, once
enterprises’ confidence in systems considered the loftiest benchmark for machine
intelligence, is effectively broken every day by
and the way they think about trust. conversations people have with Large Language
Model (LLM)-backed customer service bots and
The rush of an AI race is undeniable. sales agents. Today’s AI models have shrugged off
the deep but specific and linear approaches of the
We’ve seen it before. In 1997, Garry Kasparov past and are demonstrating more autonomy than
lost a six-game chess match against IBM’s Deep ever—in how they learn, approach tasks, and in
Blue.1 It was the first time a computer ever beat what they ultimately can do. And they’re bringing
a chess grandmaster, after decades of testing this autonomy to work, where 75% of knowledge
machine capability against humans using this workers report using generative AI; to how we
game. The victory set off a storm of excitement interact with technology, as a coding copilot and
and questions about AI and the future. Now, a by expanding voice assistant capabilities; and to
new race is underway. Many companies building nearly everything else, from robotics, to cars, to
today’s cutting-edge AI models have their sights set health care.4,5,6,7,8,9,10 Highly capable advanced AI
Technology Vision 2025 | AI: A Declaration of Autonomy Introduction 5

is diffusing across every dimension of our lives, What will the world look like as wide proliferation industries, it may look like the common framework society will uplift the world to the next level of
instantly accessible, and—effectively—always there. brings it to every dimension of our lives? It seems and communications protocol between companies capability, performance, and progress. It will spur
inevitable that as leaders start to combine their in an industry, or engines codifying the grand an evolution towards a world enhanced at all levels
This is the real disruption to focus on. Because right AI generalization efforts, they will soon enhance challenges that shape an industry—models that by AI cognition, and generate an unprecedented
now, even as executives race to implement this new and empower individuals, drive and help operate will help grow our understanding of things like wave of autonomy that will reshape technology and
generation of AI, few are looking past the separate enterprises, radically reshape industries and even physics, genetics, movement, and more. And for businesses as we know them.
pieces to truly understand the scope of what they uplift nation states. countries and governments, it brings together
are actually building: AI “cognitive digital brains” the unique knowledge, language, culture, laws, A first thought may be that this is exclusively a
that will completely reshape the role technology Take Insilico Medicine, a pharmaceutical company, and security to help industries, companies, and transition from automation using AI to autonomy
plays across the enterprise and people’s lives. which used generative AI to go from discovery citizens engage. Critically, these cognitive digital in digital systems. It’s not wrong, but it’s only part
to phase one trials of a drug in under 30 months, brains won’t operate in silos. When they begin to of the story—AI is powering autonomy in dozens
What leaders have to fully grasp is that the singularly around half the time it usually takes.11 They used interact at all levels, they will create a rising tide of of ways. It’s giving people access to skills they
most important feature of AI is its ability to learn. one model fine tuned on omics and clinical data intelligence that elevates the capabilities of every wouldn’t otherwise possess, letting them act
When AI becomes generalized, and as enterprises to identify potential targets for drug therapy. To party involved. with more initiative and less friction than before.
diffuse it across the business and people adopt it develop possible drug compositions, they used a It’s giving robots a new degree of context and
into their lives, it has the potential to become much generative chemistry engine that consisted of 500 This is why it is a “declaration of autonomy.” reasoning about the world, allowing them to take
more than just the new features and capabilities it predictive and pre-trained models. For Insilico, AI is We may call them different things, but across the on a wider and more complex range of tasks and,
provides. Enterprises aren’t merely empowering the at the very heart of what they do—shaping the very range the evolution is the same: the prolifieration most importantly, co-mingle with humans like never
workforce, creating a new channel for customer business and industry around it. of autonomous AI systems happening across before. And of course, agentic and multi-agent AI
service, or automating parts of their operations.
They are taking a technology that comes with broad A Cognitive Digital Brain at Every Level
general knowledge and is intrinsically defined by
its ability to learn and they are teaching it about
parts of the business. And when people use
It can be difficult to see this trend; at every layer of
scale it manifests slightly differently. But across the
What makes a Cognitive Digital Brain?
it, they’re further teaching it about their likes, board, this next stage for AI will infuse enhanced The cognitive digital brain will become the central nervous system for enterprise decision-making
preferences, and needs. capability and increased autonomy into everything it and continuous learning. Used to power enterprises' future ambitions, like intention-based
touches. For individuals, the cognitive digital brain architectures, it is comprised of four interconnected layers that together organize, process,
If built intentionally, enterprises can take all the will operate as a co-pilot or sidekick, something that and act on information.
distributed AI efforts they are pursuing and build will understand their job, learn their preferences,
a cognitive digital brain. They can hard-code and get to know them through its interactions, in Knowledge: Technologies like knowledge graphs and vector databases gather, organize, and
workflows, institutional knowledge, value chains, service of helping them be an enhanced version structure data from across the enterprise and beyond.
social interactions, and so much other crucial data of themselves. For businesses, it might seem
about businesses and the world into a system that more like a central nervous system—an evolution Models: Large-scale generative AI models as well as classical ML and deep learning models
can understand—and increasingly act—at a higher of the enterprise architecture into something perform critical thinking and reasoning functions to turn data into actionable outcomes.
level than ever before. that can capture the collective knowledge of the
business, its unique differentiators, and its culture Agents: Designed to be problem-solvers, tackle tasks with minimal human input, and learn
What can a person do with this power? What can and persona, and become a key orchestrator (and and grow over time, AI agents bring planning, reflection, and adaptability to the mix.
a business do deploying it across the workforce? even autonomous operator) for parts of it. For
Architecture: A comprehensive backbone is what turns AI experiments into enterprise-grade
solutions. It scales intelligence across the organization and into existing workflows and enables
repeatability, so solutions can be made once and reused.
Technology Vision 2025 | AI: A Declaration of Autonomy Introduction 6

systems are starting to take on entire workflows or AI lies in how enterprise leaders choose to harness Until now, technology systems have been broadly Because it is impacted. To start—enterprises need
customer interactions without the need for constant the new dimensions of autonomy it enables. But rules-based. Though these systems are less to realize that with growing autonomy in their
human intervention, while maintaining strategic succeeding in this new world and making the right intelligent, they are highly predictable and thus technology systems, they need to think differently
oversight. Tapping into this autonomy will stretch choices will be no small task. Intrinsically married to more trustable. As a result, their adoption and about how much they trust these systems and
the limits on what businesses thought possible. the idea of autonomy is an underpinning of trust— diffusion across enterprises is widespread. So now, what guardrails they may need to impose. Sakana
Accenture research has found that with its ability to and for enterprises, it’s trust that will be the biggest as we look ahead to a world that will be defined AI, an AI research firm, perfectly demonstrated
reimagine and augment complex tasks, generative backstop to tomorrow’s growth. by technology systems that both have and create why while testing their new system called “The AI
AI is expected to drive productivity gains of 20% in greater autonomy, we’re looking at a future where Scientist.”14 The system autonomously conducts
companies leading in AI adoption.12 The Only Limit is Trust trust is the most important differentiator and scientific research using LLMs, and in one run, was
the determining factor to AI diffusion within an given a problem it couldn’t complete within the
What we have today is the spark for unbounded Think about how trust defines the human organization. After all, we can only let systems be as experiment’s set time limit, so adjusted its own code
growth and innovation—as well as disruption. As experience—the relationship between a parent autonomous as we trust them. to give itself more time. Sakana AI has pointed to
ever-greater autonomy reduces friction within and child, for instance. We surround babies with this act as creative, but also demonstrative of the
and between organizations, letting us get more guardrails. From literal ones, like those in a crib, But the ramifications of this are not as obvious fact that an AI model with the ability to bypass a set
done faster, early movers will be able to secure to more figurative ones like cutting up food or as you might think. Of course, most leaders will constraint has major implications for AI safety.
advantages that last decades. Failing to act or covering sharp corners around the house. As they be well versed on how bad actors can spread
waiting too long will give ground for competitors, grow up, we learn to trust them more. They don’t misinformation more effectively through deepfakes And beyond an enterprise's trust in the AI models or
new and old, to disrupt industry norms just as need to hold your hand to cross the street, but we or conduct more convincing phishing attacks with systems it uses, growing autonomy is also disrupting
we saw in the digital era. And consider this: Less still walk next to them. They can play outside by better emails or spoofed voices of real people. the trust enterprises have built with people, in a lot
than 1% of today’s global internet market cap themselves, but only inside the fence. The more Or how biased decision-making can rear its head of different ways.
was founded in the first two years after Netscape our trust grows, the wider we paint the boundaries even with AI. To be clear, these are real issues, with
Navigator generalized the internet for the world.13 of the guardrails. Until, one day, they are fully ever-growing efforts for content watermarking Take the same synthetic content that bad actors
Now, it’s been a little over two years since formed adults. We’ll still check in—but they’re their or deepfake detection tools urgently seeking use; many enterprises are using the same core
ChatGPT’s release. Our foray into this generation of own person now, with the autonomy to make their resolutions to them. But this narrative pins the AI technology to great effect. AI-generated marketing
AI has only just begun, and with such large stakes, own decisions. trust conversation exclusively on bad actors and materials, chatbot conversations, product
it’s vital enterprises start now before they’re left exploitation. That’s simply not the whole story. To recommendations—the use cases are ever-growing.
irreparably behind. Critically, this example demonstrates how trust achieve true autonomy—in systems, throughout the But what happens when a customer finds out that
and autonomy are inextricably linked. But it also workforce, and with customers—leaders need to a product photo was AI-generated? Or if they
demonstrates the nuance of trust leaders now need think about trust more holistically. Like the analogy believed they were speaking to a customer service
To understand more about how digital to consider. The relationship between parents and of guiding a child into adulthood, trust is about representative, only to learn it was an AI agent all
platforms, data & AI, and digital foundations children hinges on both emotional and cognitive the confidence one develops in AI to perform as along? These interactions could leave customers
are empowering enterprises to grow through components of trust. Guardrails help foster a loving, intended from all dimensions—policies, morals, feeling duped by the business.
change and disruption, please see our work on nurturing, and safe environment, but also help ethics, and emotions—so that one can let it perform
Reinventing with the Digital Core parents build their own confidence in the child’s in a state of autonomy. Which means, trust isn’t just Or look at AI in the workforce. Every day more
decision making ability. We don’t really need to about when AI is taken advantage of, but the harder workers are finding value in using AI in their
differentiate between these two dimensions when question of how trust is being impacted even when jobs; in May 2024, over 40% of users had started
We are at the start of so many possible paths it comes to people, but with technology they we are using AI exactly as we intended. incorporating it just in the past six months.15 But
forward. The key to accessing the full potential of are different challenges with different solutions. they are hiding this from their employers: more than
Technology Vision 2025 | AI: A Declaration of Autonomy Introduction 7

half of workers using AI are reluctant to admit it and relevance that only AI agents can provide at scale. require dedicated teams of domain and decision Many companies are already familiar with ideas like
worry that using it for important tasks makes them But that autonomy needs to be facilitated by trust. scientists (or AI Ops teams) who will constantly explainability, transparency around data collection,
look replaceable. This isn’t a question of how much How much customers trust the enterprise, or test, evaluate and build the accuracy, predictability, debiasing, and other maturing techniques, but as
do workers trust the AI they’re using, rather it is enterprise leaders trust their systems, or workers consistency, and explainability necessary to leaders look to scale their use of AI, these efforts
proof that AI is shaking up the trusted relationship trust their employers, how much people trust AI, or maintain cognitive trust in the system. This is new will become a critical bridge between technical
between people and their employer. Employees dozens of other permutations across the ecosystem territory, with no one-size-fits-all solution. Every solutioning and the humans interested and invested
are used to having well-developed career paths, of relationships an enterprise has. enterprise has their own trust-building moments, in using the technology. Questions will swirl around
defined roles, skill expectations, and a shared technologies, AI strategies, and key relationships to how an AI is trained, who it is working for, and how
understanding of how work performance translates This is why trust is not simply one of many trends focus on. But broadly, any path forward will center it makes decisions—these are unavoidable. But what
into job stability. The infusion of AI is bringing in this year’s report. It is not a consideration for on addressing trust across systems and data, AI enterprises can do is be prepared with an answer,
uncertainty to this. businesses—it is the consideration. With every itself, and people. which is why they can’t sit back and wait, and need
company beginning to reinvent themselves with the to make responsible AI a key part of their strategy now.
For enterprises, trust is a crucial currency generalization of AI, the technology itself cannot be First, enterprises need to bolster the cybersecurity
underpinning their relationships with customers, the only focus. Reaping the benefits of AI will only and trust of their digital systems. The good news Lastly, the third and uncharted part of the roadmap
employees, regulators, and shareholders. Until now, be possible when it’s built on a foundation of trust, is that for systems and data, building this new is finding a new path to people-driven trust. We
this trust was built in small moments—moments and this needs to be every leader’s first priority. foundation doesn’t mean starting from scratch. know where we need to get to—new touchpoints
that AI is changing. Think of the micro-interactions Many previous technology and strategy investments and ways to establish and maintain trust with people
happening across businesses every day. A great Pathways to a Stronger Foundation can pay new dividends now. Cybersecurity as the generalization of AI disrupts traditional
sales rep saving customers money, or a support strategies like zero trust and entity behavior interactions. But how to get there will be different
rep going above and beyond to solve customers’ Trust isn’t gone in the world of AI—but it is analytics, for instance, will be critical. You can’t for every organization, so the place to start is with
problems. Quality service from a practitioner becoming far more dynamic and essential to control bad actors, but you can control how you self-directed questions: What will career paths look
or provider. Calling a customer for identity enterprise plans. With AI, enterprise leaders are protect systems and people from them—and with like when many entry-level jobs can be done by AI?
confirmation. On-time delivery of products. Every going to need to navigate both the emotional AI’s dependence on data, protecting everyone’s What will establish job security for employees who
one of these moments can, and will, be disrupted and cognitive dimensions of trust. The emotional data is increasingly important. Distributed ledger leverage AI to streamline their work? How will we
with AI. And many will be better for it—rich with dimensions—do people love AI, fear it, think it is technologies that foster ecosystem-scale trust maintain a personal touch with customers if our
more autonomy, less friction, and better outcomes. aligned with their interests, or feel taken advantage are also a great example of adapting traditional “frontline” support is AI agents? Enterprises should
But how far can you go before trust becomes an by it—are often considered publicly, but will need networks of trust toward new, technology-based seek to answer these questions in ways that will
issue? How will you reinvigorate the critical human real policy and governance as enterprises seek to ones. You don’t have to trust the entities using promote the potential of the symbiotic relationship
moments that build it? further diffuse the technology. And these efforts these technologies, because the system ensures between people and AI. Whether it is educators and
will only be successful with complementary action they comply with whatever agreements are put in students, mentors and proteges, or superheroes
These are the issues that leaders need to be taken to address the cognitive dimensions of trust: place. Ultimately, great cybersecurity overall will be and sidekicks, the world is filled with mutually-
tackling. Autonomy is the key to the next generation Does a system act reliably, competently, can it paramount to achieving AI trust and security at all. beneficial teaching and learning relationships that
of business growth and innovation. We want navigate challenges and still act as expected, within should inspire our future with AI.
employees to be able to work more autonomously, the guardrails laid out for it. This is a key aspect for The second dimension of the roadmap is thinking
with a fleet of agents at their command. We any system operating with autonomy, and especially about building trust in AI itself. By now the field
want customers to be able to freely interact with with AI where it is, by nature, one that will be relied of responsible AI is becoming an established
autonomous enterprise systems, purchasing on on to learn, grow, and act based on intent, not discipline, and one enterprises will increasingly rely
demand or enjoying a level of customization and necessarily explicit direction. Supporting this will on as they look to ethically steward their strategies.
Technology Vision 2025 | AI: A Declaration of Autonomy Introduction 8

The 2025 With a firm and clear approach towards building

trust in AI systems, and by actively building the
cognitive digital brains that will become an essential
part of the enterprise DNA. The result will be a
Finally, The New Learning Loop explores the
impact of cognitive digital brains through the
Technology Vision cognitive digital brains that will create scaled
intelligence across the spectrum of society,
dramatic increase in technology diffusion touching
every walk of business, consumer, and societal
most valuable reinvention engine you have:
your people. Employees are starting to bring
Trends businesses will be able to unlock the limitless interactions. It sets the stage for the emerging AI AI to work, and employers know the power it
potential of AI today. It seems prudent to mention era, where we will rapidly expand digital ecosystems can wield. But we need to change our mentality
that in 25 years of producing the Technology Vision, and increasingly trust autonomous systems to find from automation to equipping your people with
few technologies have had the widespread impact new ways to innovate with us. the power to automate—giving them the tools
on business, industry, and even technology itself to innovate, reimagine new ways to do things
that AI is poised to have now. We anticipate we are Your Face, in the Future pushes the thinking and drive progress from the ground up. We are
living in a time on par with the biggest moments in further, asking a simple yet critical question: if the building a virtuous cycle where people teach
technology, one which will be shaped and defined world is shaped by AI and increased autonomy, and learn from AI machines, and AI machines do
by AI-powered autonomy and the emergence of AI- and brings super-human consistency to everything the same with people. A cycle that will let both
based cognitive digital brains at all levels of society, it touches, where does that leave your brand and unleash new levels of performance and diffusion,
and we’ve only just begun. unique enterprise personality? Enterprises are at all underpinned by trust built through ownership.
the crossroads of an intermediation challenge. They
In the interest of preparing business leaders for the have the opportunity to radically transform the way What is the world going to look like in 30 years?
transformative journey ahead, the Technology Vision they engage customers and improve the relevance Around the time of Kasparov’s groundbreaking
this year is a deep investigation into this declaration of their customer journeys, but to do so, they must match, laptops were just starting to become
of autonomy. Our trends explore the business realize that their company’s AI personality is as popular among business workers, no one had ever
transformations—and trust revolution—that will critically important as its traditional brand built over heard of an iPhone, and economist Paul Krugman
happen as generative AI ripples across dimensions time by small, personal human interactions. infamously declared the internet would prove
of customer experience, technology development, to be of no more value than the fax machine.16
the physical world, and the workforce. When LLMs get their Bodies explores the Now, we see the beginnings of a future where
manifestation of AI autonomy in the real world, AI cognitive digital brains fuel every layer of
The Binary Big Bang tracks the emergence of and how a cognitive digital brain can transform an society, interact with each other, and bring new
language models coupled with agentic systems, and enterprise's physical presence. We are reaching a intelligence to everything and more autonomy to
how they challenge conventions around building watershed moment as the power of generative AI is everyone. And the question enterprises need to
software and crafting new digital ecosystems. This applied to physics and the field of robotics. Gone ask themselves is not whether this will pan out, but
is a redefining moment in the world of software are the days of narrow, task-specific robots that rather how they will invest in fostering the trust
engineering, where the role of programmers has require specialized training. A new generation of needed to make this future a reality, and what they
largely remained the same since Ada Lovelace wrote highly tuned robots with real world autonomy that will be able to do with this limitless capability.
the first algorithms for Charles Babbage’s Analytical can interact with anyone, take on a wide variety of
Engine. The trend dives into a generational tasks, and reason about the world around them will
transition, as leaders rethink how digital systems expand robotic use cases and domains dramatically.
are designed—building the foundation for the
Technology Vision 2025 | AI: A Declaration of Autonomy 9

01

The Binary
Big Bang
When AI expands
exponentially, systems
are upended

Organizations are entering a generation-defining

moment of transition: the Binary Big Bang. When
foundation models cracked the natural language
barrier, they kickstarted a shift in our technology
systems: how we design them, use them, and
how they operate. They are pushing the limits of
software and programming, multiplying companies’
digital output, and laying the foundation for
cognitive digital brains that infuse AI deeply into
enterprises’ DNA. We are now at the precipice of
more abundance, abstraction, and autonomy in
our technology systems than ever before, and the
decisions enterprises make today will profoundly
impact what they can achieve for the next decade.
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 10

1843 1946 1959 1973 1983 1995 2006

Ada Lovelace writes Researchers unveil COBOL, a major Alto, a computer with a Microsoft releases Sun Microsystems Amazon launches
“Note G,” widely ENIAC, the first programming language graphical user interface, is Microsoft Word21 releases Java 1.0 as a AWS, and 12,000
considered to be the general-purpose for business, finance, developed at Xerox PARC20 ‘write once, run anywhere’ developers reportedly
first complex computer electronic computer18 and administrative programming language22 sign up on the
algorithm, on a sketch systems, is created19 first day23
of Charles Babbage’s
Analytical Engine17

2008 2017 2023 Future 2026 2028 2029

Apple opens the “Attention Is All You Toolformer, an LLM that People regularly use A major ERP company An agent-powered
App Store24 Need” paper proposes the can teach itself to use personal agents to access launches a new agent-first startup goes public,
Transformer architecture external tools through data and complete tasks platform design that lets and has the highest
and paves the way for APIs, is created26 across digital ecosystems users create custom market cap-to-employee
modern LLMs25 data dashboards using ratio in history
natural language

2030 2032

Agents, not people, Interacting with agents

are the primary users of surpasses apps in average
most enterprises’ internal consumer time spent on
digital systems smart devices
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 11

The Big
shift in technology’s building blocks. Where
software has long been a tool for enterprises to

Picture
create new capabilities and outcomes, AI will
multiply the output, pushing limits on who can
develop it and how it can be used. It’s why we call
it the Binary Big Bang. A new technology paradigm
is on the rise—defined by abundance, abstraction,
and autonomy—and it’s about to drive a massive
In September 2024, Marc Benioff announced that expansion of businesses’ digital footprints.
Salesforce would “hard pivot” to Agentforce, a
platform for building and deploying autonomous Enterprises will soon operate in a technology
AI agents.27 It’s rare for a company of such scale to landscape where new digital concepts come to life
pivot like this. But Salesforce realized something
in hours instead of weeks. Where they can end one-
groundbreaking and formidable that every
company needs to recognize too: size-fits-all digital solutions, as the drastically faster
rate of creating digital systems and connecting them
We have just entered the Binary Big Bang. together enables a level of personalization never
before possible. It will be a new era of customizable
It’s a generation-defining moment, and specifically, technology where unbridled digital experimentation
a moment of transition. When foundation models and innovation will become not only possible, but
first emerged, they cracked the language barrier a key feature of business growth strategies.
between people and technology. And while
these natural language capabilities are well This is the future we started moving towards the
known—driving today’s innovations in AI chatbots, moment AI broke the language barrier—but we’re
summarization and synthesization, creative not there yet. First, there’s a moment of transition
generators, and, perhaps most consequentially, that leaders need to use to set themselves up for
coding assistants and agentic workflows—their full tomorrow’s paradigm and their next decade of
breadth of impact is poised to go much deeper. digital growth. And that’s where we are today. It’s
why companies like Salesforce—which defined SaaS
These AI models have the potential to do more and introduced software standardization—are now
than deliver automation on top of existing business taking such substantial action. With Agentforce,
processes; they’ll soon be creating new processes, they’re leaning into tomorrow, letting every
workflows, and software. This is a groundbreaking company create their own, unique, personalized
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 12

experiences. And other savvy businesses are embedded in operating systems that enable transition phase than those which are static and
recognizing that they also need to look at the previously impossible features like Recall, which waiting for periodic indicators to initiate change. It’s
bigger picture—beyond the immediate AI use reminds users what files they were working on or time to stop thinking about technology as a tool and
cases (valuable as they are) and to the deeper shift websites they were browsing even months later; see it as your biggest competitive differentiation.
happening in technology that will set up their AI and true agentic systems like AutoGen, which is a
futures. framework for multi-agent interaction laying the Of course, recognizing the opportunity we have
groundwork for agent-to-agent automation.29,30,31 today is only the first step. There is no defined end-
Agents are a key demonstration of this perspective. Other hyperscalers are putting their weight into state yet or proven best practices for tomorrow’s
Right now, they represent the pinnacle of AI this trend as well. AWS’s Amazon Q Developer new technology paradigm. And while agents are
innovation, and as such, many companies are racing and Google’s Gemini Code Assist are key tools the best new architecture today, we expect more
to leverage agents in their products and services. that will radically accelerate digital systems new architectures to emerge in the future. But
But if leaders only look at agents for how they development.32,33 There are also novel innovations of what we do know is that three forces—abundance,
augment a piece of software, and not how they are parallel (or background) programming coming into abstraction, and autonomy—are emerging as the
challenging the nature of software itself, they will be play, like with Cognition AI’s ‘Devin’ code assistant pillars of tomorrow’s technology.
overlooking key opportunities and disruptions that that takes instruction from the project manager and
define this transition period. performs activities in the background.34
And as enterprises step into this transition and
Take Adobe, which introduced several AI features Think back to how early digital movers like Amazon start to build new strategies, they will need to
built on its Firefly generative AI models.28 While they and Netflix didn’t just see the use of social, mobile,
still require human input, these features let users analytics and cloud technologies as a way to deeply understand these pillars, what they mean
create and edit images through natural language augment a business—they saw how it could be for technology, and the ways they drive business.
alone, easily accomplishing feats that would leveraged to upend the very value proposition of
otherwise have required extensive knowledge their industries. Now today’s storm of foundation-
of Adobe’s apps. And these aren’t just any new model innovation and its ability to create more Starting with abundance: As we’ve seen through
features. Effectively, by introducing generative AI technologies and solutions on the fly is showing us coding copilots, the creation of digital systems
components, Adobe is transforming apps designed that our base assumptions about digital technology is getting a lot cheaper and faster. Amazon’s
as a toolbox for creatives, to apps with a creator are becoming obsolete—and the chance to upend generative AI assistant for software development
in them. As the capabilities improve, it begs the how businesses compete is here again. was able to save the equivalent of 4,500 developer-
question: do most end users need to access the full years of work when updating an application to
app, or can they—or even their agent—just talk to This is why it is critical leaders recognize the Java 17.35 And even the secondary effects of the AI
the creator instead? Binary Big Bang for what it is: a brief moment of race are having an impact—Jensen Huang stated
transition where enterprises can take stock of the that NVIDIA has driven down the marginal cost of
This is only one of many revolutionary ways changing technology landscape and carve out how computing by 100,000x.36 What’s happening is, one
agents are challenging our existing technology they compete tomorrow. Companies that lean into of the most complex things a company can do—
conventions. Microsoft is exploring natural this moment and lay the foundation for systems that generate code—is becoming more efficient than
language-based agents through myriad other can autonomously create new solutions, identify ever. And this doesn’t just impact the bottom line.
approaches: coding co-pilots that are transforming malfunctioning (or successful!) features, and It’s a huge opportunity to accelerate modernization,
standards for developer efficiency; agents even self-heal, will cover far more ground in this tackle backlogs of projects, and more. But it also
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 13

means competitors will be moving just as fast and defined purpose. Autonomy breaks us out of that, purpose architecture are immense, but not without
new disruptors can emerge with even less friction. as systems that can build and execute code on their challenges. There will no longer be a reference
own are poised to become powerful orchestrators point for how an architecture should look. How we
Second, if speed and efficiency will expand the and operators of the business. The opportunities govern, guardrail, and benchmark systems has no
proliferation of technology, abstraction will expand are immense; it suggests a future of frictionless and blueprint anymore. Every system will be unique in its
who and how we use it. Just like COBOL did for intent-based systems. But it also means a radically own right, so what will good look like for you?
assembly languages and graphical user interfaces different way of building and training those systems. But uncertainty isn’t cause for inaction. There
(GUIs) were to command-line interfaces, modern AI To quote Geoffrey Hinton, the godfather of AI, “We is already no shortage of agents, copilots, and
systems will be a new bridge between people and designed the learning algorithm…but when this platforms giving access to this powerful new
machines. Adept at both code and natural language, algorithm then interacts with data, it produces technology, and the race to integrate it has begun.
agentic systems will let people shift from operating complicated neural networks that are good at doing This is why it remains essential that leaders see the
software themselves, to making requests that things, but we don’t really understand exactly how bigger picture. Today’s technologies are not the
kickstart an agentic process that reasons through they do those things.”37 See Reinventing with a endgame—they are how we are getting there.
the request; considers functions, APIs, data, and Digital Core: Chapter 2. A technology future of abundance, abstraction, and
tools to tackle it; and then creates and implements autonomy is what leaders need to be planning for.
a plan that can be packaged and reused indefinitely. Autonomy means we might not know how, or
Adobe Firefly is an example of how leveraging this why, a system makes particular decisions—but
Every company must be prepared to forge a new
abstraction to reinvent how software is designed— we have to raise them to make the right ones.
and for what users—can be a key wellspring of Trust will be essential to navigating the Binary Big technological footprint, founded on AI—a unique
opportunity. But it’s also much more than just ease Bang. Businesses are used to being able to buy DNA identifying and differentiating them as they
of use. To effectively take advantage of this modern software off the shelf, not needing to caretake its
launch into tomorrow’s technology landscape.
coding paradigm and increase adoption while decision-making almost the way you would with a
maintaining structure, this new layer of abstraction child. Businesses will need more than the technical The Binary Big Bang is the moment to get your
will be a necessity. Like providing APIs and SDKs for infrastructure to train and develop these models—it footing—but it won’t last long. How can you take
developers, in a world of autonomous programming, is more crucial than ever to also have guardrails and
advantage of it today?
an abstracted interface between the architecture reinforcement in place to ensure good outcomes.
and business workflows will be required to avoid Creating the internal policies that will allow
breaking other elements of the system. While today enterprises to trust this technology in the hands of
we are seeing the abstraction in packaged software their people, the governance to ensure autonomous
or discrete applications, eventually enterprises decision-making is congruent with enterprise
will need this layer of abstraction across the entire objectives, and the cybersecurity and testing
organization to generate the speed and governance practices to keep systems secure and resilient will
necessary for tomorrow’s enterprise. (See be crucial to navigating the new waters enterprises
Implications for more.) are wading into.

The third pillar is autonomy, which is poised to The Binary Big Bang will feel hectic. The
be the most significant. Today’s architectures are opportunities for unprecedented customization,
designed against executing a singular and rigidly- and the advantages of having a completely fit for
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 14

The Technology

Driving the Binary Big Bang is a set of technologies

reshaping how we think about computers, software, and
digital systems. Understanding these components will
help organizations set themselves up to rapidly grow
their digital footprints and lead in tomorrow’s technology
landscape of abundance, abstraction, and autonomy.

• Agentic Systems

• Digital Core

• Generative UI
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 15

Agentic Systems
What is it? Who is doing it today? Research interest in agentic systems is surging
Over the past two years, researchers have discovered powerful ways to extend
AI agents and agentic systems offer a powerful Agentic systems show great promise with small the capabilities of LLMs to create AI agents and agentic systems. Companies
way to leverage LLMs and other foundation models pieces of code, and given documentation and are transforming these research efforts into commercial offerings rapidly.
to complete complicated and compound tasks. examples, can call functions and APIs with high
Agentic systems take the power of language accuracy. Research has also demonstrated they can Number of research papers relating to Agentic Systems, 2020 - 2024*
models and extend them by integrating them with create functions and APIs to use later. Companies
methods for reflection, tool use, planning, and are rapidly integrating these capabilities into
collaboration. These methods transform the models new models to accelerate engineering velocity.
from simple prompt and generate functions into One of the leading agentic systems for software
reasoning engines that can tackle a huge variety of engineering today is Anthropic’s Claude 3.5 Sonnet. 1,800
challenges. When tested on SWE-Bench Verified, a software 1,576
1,600
engineering benchmark of real-world issues from

How does it connect to the trend?

GitHub, it achieved a remarkable 49% resolved
rate.38 In 2023, agents had a rate of less than 5%.39
1,400
17x
1,200
Agentic systems may eventually be one of many
architectures or models that deliver abundance, 1,000
abstraction, or autonomy—but right now they
are the one capturing the most attention across 800
researchers and the technology industry. They are 588
600
already helping to write code, and as these systems
improve, with better reasoning capabilities, larger
400
context windows, and validation methods, they will
be able to take on an even greater set of challenges. 200 50
50 71
71 91

0
2020 2021 2022 2023 2024*

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on ArXiv papers; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 16

Digital Core
What is it? Who is doing it today? Agents in the Digital Core
Organizations are planning to integrate AI agents into the heart of their
The digital core is the critical technological Today’s agentic systems can’t build and maintain digital systems—not just for accessing information but for upgrading
capability that can create and empower an the entire digital core—but they’re tackling pieces functions and building new components of their digital business.
organization's unique reinvention ambitions.40 of it. Salesforce’s Agentforce acts as an agentic
Key to this is a composable architecture that system across many parts of the Salesforce When do you estimate your organization would enable the following
emphasizes modularity and interoperability. platform.41 On the surface, it might just seem like capabilities for AI agents to integrate with your digital architecture?
Composability relies on independent self-contained a bundling of common generative AI features, but
components that can be connected to build high- it can also make plans and leverage existing flows
level functions and applications. They can come and Apex code to take actions within the Salesforce
Upgrade and modernize
from internal systems, PaaS and SaaS providers, and platform. functions and integrations 48% 52%
other external parties. But in all cases, they need to
Assure the quality of digital
operate independently, be trusted and verifiable, functions and systems 46% 54%
and be discoverable and usable by composers.
Access functions from
internal systems 45% 55%
Build automations that tackle
How does it connect to the trend? specific tasks
45% 55%
Access data from across
As agents and agentic systems emerge, they’ll the organization 38% 62%
bring new urgency to composability. True agent
autonomy will only be achieved if companies Build new functions 33% 67%
seamlessly connect data sources (both internal and
Build workflow automations that
external) with analytical platforms that can use that span across the organization 32% 68%
data to drive decision-making and useful actions.
Agents will, in other words, be major users of the Access functions 3rd party systems 29% 71%
digital core, leveraging it to answer user requests
and compose new digital solutions. And further on, Near term (within the next 3 years) Mid or long term (4 or more years)
they could even play a role in building, validating,
and updating composable components.

Source: Accenture Technology Vision 2025 Executive Survey, N=4,021

Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 17

Generative UI
What is it? Who is doing it today? Redefining our relationship with computing
The GUI has been the predominant method of human-machine interaction
Generative UI is the concept of leveraging AI In one early example, Google demonstrated its for many decades, but executives believe this is changing as agents become
techniques to generate user interfaces, commonly Gemini model generating interface elements to available. This shift will involve natural language but also the ability for systems
with the goal of offering a personalized experience. help plan a birthday party.42 Instead of responding to generate and compose UI elements dynamically to meet a user's request.
The long-term vision is a system that dynamically in a bulleted text list, Gemini generated an
generates interfaces in real-time, based on an interactive list of party themes, each with an In your estimation, when do you think the general public will use AI agents
individual’s context and needs. And this is not just image, description, activities, and food options. more frequently than apps or websites to complete day-to-day digital tasks?
populating a predefined layout with user-specific Vercel is also hoping to accelerate this transition
content, but completely changing the structure, to generative UI with its AI SDK 3.0.43 This update
45%
flow, and interaction methods. These capabilities lets apps use React Server Components to stream
can be applied to websites, applications, or even UI components directly from LLMs, without heavy 39%
40%
agentic systems, and could rely on text, voice, or client-side JavaScript, enabling a much more 37%
other intuitive interactions. engaging way for LLMs to respond to user requests. 35%

30%
How does it connect to the trend?
25%
Agentic systems and composable digital cores
will radically transform how we think of the user 20%
interface and unravel a core component of 15%
today’s software paradigm. For decades, the high 15%

cost of software development and the low cost

10% 8%
of software distribution have driven the idea of
creating a single UI that must work for every user.
5%
But now, as agentic systems advance and begin to 1%
take more actions on our behalf in the digital world, 0%
they’re driving a new software paradigm where
2025-2030 2031-2035 2036-2040 2041-2045 2046-2050+
cheaper code and language-first interfaces make
dynamically generated, custom UI components
increasingly feasible. Source: Accenture Technology Vision 2025 Executive Survey, N=4,021
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 18

The Implications

Leaders need to As leaders set their sights on new avenues of

opportunity generated by the Binary Big Bang, it’s
critical to analyze how the emerging paradigm of
processing, computer-aided design, or photo
editing. It wasn’t always a perfect system, but it was
better than the past and worked extremely well when
generative AI can become the cornerstone of UI.
Through simple natural language interfaces, it can
respond to user requests by kicking off a swarm of

start thinking abundance, abstraction, and autonomy is going to

impact the technologies we use today. While it won’t
our digital needs were more targeted. digital activity. Unlike apps that are limited to specific
software features, composable digital cores will

about a world be immediate, one thing leaders need to watch and

plan for is the evolution of apps.
But now, we rely on technology for so much more,
and in many ways this old paradigm has become
a bottleneck. Just look at the massive suite of
let agents autonomously create connections, build
tools, and leverage huge swaths of business data as
needed. And because these systems are designed

beyond the app It’ll be a welcome transformation for every enterprise

looking to further diffuse the capabilities of AI across
software tools that enterprises need today, and how
they create data silos and complicated workflows.
to learn, as they build new workflows, they can also
create pathways for future agent requests. In this

paradigm today.
the workforce. The farther we push into this new era Or how some software tries to overcome that siloing, way, every time an agent reasons through a problem,
of technology, the more clinging to older paradigms only to end up becoming sprawling, feature-rich it ultimately has a compounding effect on the
and architectures may end up limiting the full applications so complex that fully utilizing them company’s wider ability to tackle problems.

Agentic systems potential of new systems. takes years to master. It can end up burdening
the enterprise, from security oversights, to excess This is why leaders need to start thinking about

will be at their
To understand, it’s helpful to revisit how today’s spending on upkeep and maintenance of aging a world beyond the app paradigm today. Agentic
technology evolved. Since the 1980s, software systems, to the severe productivity impact of app systems will be at their most effective when
applications have (at a high level) been a way to fatigue when people are forced to jump between connecting components across the enterprise,

most effective package function and data together with an easy-

to-use graphical user interface (GUI). A single click
applications to accomplish a single task.45 turning them into new sources of customer value,
automated workflows, or more. This isn’t merely
on the user’s side could execute tens or hundreds of With tomorrow’s technology paradigm on the a technical question though—the evolution of apps
when connecting lines of code behind the scenes. Ultimately, software
was built for people. Some of the earliest programs
horizon, leaders are starting to disrupt the
very fundamentals of how we design software.
has far reaching implications leaders might not be
prepared for.

components across replaced “computers,” who back then were people

who performed mathematical calculations by hand.44
By leveraging the abstraction that agentic or
autonomous systems offer, they’re realizing we can How many of today’s workflows and products are
Other early systems were focused on streamlining unbundle the function/data/interface paradigm that built around end user-facing apps? There are app
the enterprise. and accelerating tedious business operations like
accounting and inventory management, leading to
has defined software for so long and develop a new
one that unlocks the power of computers differently.
interactions we get a lot of value and data from today,
and they are fundamental to many of our customer
the development of very siloed solutions. The same and employee experiences. How will the enterprise
was true when personal computers emerged— Imagine this: instead of users having to find the right react when all these things start to change?
apps were designed for specific tasks like word apps or switch between them to get tasks done,
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 19

Just look at the consumer app space: The last will change—but the disruption is upon us and
generation of consumer applications was largely enterprises need to start building tomorrow’s
designed for mass appeal—meaning that they were strategy. Because the truth is, most companies aren’t
broadly distributed, and how users could interact set up for this yet. Leaders will need to transform
with them was rigidly defined. But the abundance, both their technology and organization to keep pace
abstraction, and autonomy of tomorrow’s technology with the opportunity being generated by AI. They
systems is turning all this on its head. Enterprises need to rapidly invest in building a composable
can start imagining new ways to interact with digital core so agentic systems can bridge data
customers, like generative user interfaces, where silos and build cross-organizational solutions, and
agentic systems take personalization beyond an they need to break down employee silos. Most of all
algorithmic newsfeed or list of recommendations to they need to prepare for a world where it’s not just
a full interface designed just for an individual. You people, but machines, using their technology and
could imagine an e-commerce app where today all connecting with the business.
users see the same UI with different items listed,
but tomorrow could be shopping in completely Soon enough, the abundance, abstraction, and
personalized ways, some seeing in-depth reviews, autonomy of tomorrow’s technology will start
others discovering products through videos, or even to rapidly expand digital ecosystems, and the
an e-commerce store with no products at all—just companies ready to leap on this will quickly outpace
an agentic interface that will work with the user to the ones left catching up.
custom design their own items.

Even if this feels far out, enterprises need to be

thinking about how these changes to technology
will shape touchpoints they have with customers
and partners. Earlier in this trend we talked about
Adobe Firefly, Adobe’s family of creative generative
AI models. The company uses the models to power
their own platforms, but they’re introducing Firefly
to others as well. Slack, for instance, is integrating
Firefly alongside other agents from companies like
Salesforce, Asana, and Writer so users can access
agents without breaking their workflow.46,47 For
some enterprises, it could make sense to shift to
building functionalities for other platforms, meaning
potentially new business models.

The Binary Big Bang is still just beginning, and

we don’t know the exact end state of all that
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 20

What's Next
It’s time for businesses to prepare for the new technology paradigm of abundance, abstraction,
and autonomy. We are in a transition period today, but AI agents are already letting businesses
transform how software is designed and used, expand their digital footprints at an unprecedented
rate, and reimagine the value propositions of their industries.

If you’re an early adopter? If you’re just preparing to start? If you want to take a slower approach? How to preserve trust?

Define your new digital ecosystem: Map out ecosystem partners’ agentic offerings: Prepare your digital core for agents: Monitor autonomous systems’ activity:
In tomorrow’s technology landscape, agents will be Many platforms and software providers are integrating Every company is going to need to reinvent their digital As agent-powered autonomous systems start to play
critical to digital ecosystems. They will be users and agentic systems into their existing offerings, making it core to focus on composable integration. If you haven’t a bigger role in digital ecosystems, companies will
definers of your digital core, continuously building easy for companies to start leveraging them. Partners started yet, do so immediately. A modular digital need to closely surveil them and ensure guardrails
functions, QA systems, and more, forcing companies and providers will help take data already stored within core will be critical for AI agents to leverage data and are in place. What data are these systems accessing,
to rethink their architectures. It’s a big, ecosystem- platforms and begin to architect agents across that functions across your organization in the coming years. who is directing them, what is the quality of their
level change, and early adopters have the chance platform. Scan who in your current digital ecosystem is And even if you aren’t using agents yet, a strengthened outputs, and more. Transparency here will help to
now to start tackling some of the big questions that pivoting to agents and where that is poised to disrupt digital core will accelerate your other digital efforts. increase employees’ trust in the systems. As you
come with it: How do you track and steer millions of or create opportunity for you. Be open to creating create a monitoring system, lay out a governance
autonomous agents? How do you embrace abstraction custom agents of your own. Watch signals to predict upcoming industry impacts: and technological roadmap for implementation. Also
in your technological foundations to redefine roles and Today’s transition period will be brief and high develop communication and maintenance plans so
processes? How do you manage exponentially growing Start experimenting with agents internally: stakes, so those taking a slower approach will need your organization understands how the monitoring
digital systems? Companies that are familiar with building LLM-based to know when to stop waiting and jump in. Since works and your guardrails keep up with advances.
systems, and want to make a bigger investment in the generative AI applications are more defined than
Identify the highest value opportunities Binary Big Bang, can start by looking for opportunities agentic systems, many will see these more concrete Train autonomous systems to make good decisions:
in tomorrow’s technology landscape: internally. A good way to begin is to create task-specific applications as a valuable on-ramp to gain experience. AI agents are amazing technical feats but are by no
Track research in agentic systems to inform your art of internal agents, using the principles of abundance, But it’s important not to get too comfortable. Work means perfect. They are computationally expensive,
the possible. These systems are already driving greater abstraction, and autonomy to guide how you choose with industry partners to form a strategic roadmap non-deterministic, and can lack explainability. But
abundance, abstraction, and autonomy in technology use cases. After starting small, you can move to ensure you’re ready when the moment strikes. just as retrieval augmented generation (RAG) can
than ever before—and now is the time to discover modularly, over time expanding the functions and data ground an LLM, so can code and functions ground
where they can meet your business needs best. Detail your internal agents can access and using them to an agent, making them more explainable and
the data and technological infrastructure required learn and prepare for building external-facing agents increasing trust in them. For companies using agents
to enable an agentic solution. Set your goals for the in the future. today, determine the feedback loops rewarding and
solutions, how you will measure progress against them, reinforcing the right agentic outcomes. And for those
and evaluate them continuously to figure out your taking a slower approach, work with an industry
optimal agentic architecture. partner now to identify relevant, high quality training
sets or your most important proprietary data to lay
the groundwork for when you are ready to start.
Technology Vision 2025 | AI: A Declaration of Autonomy The Binary Big Bang 21

A Portrait of the Future

Cal works in operations for a national pizza chain. occupancy information while others don’t, so it
One morning over coffee, they’re listening to requests permission to find satellite data from
a colleague complain about how boring it was another agent to fill those gaps. Cal okays the
charging their car at a station last night. An idea move, and the agent works with a satellite imagery
sparks: What if bored—hungry—people could company’s agent to determine each station’s
order pizza to their cars while charging? How occupancy over the last six months. Then it logs all
many EV charging stations don’t have easy access EV data and combines it with local pizza sales data,
food nearby? It could open a whole new market which it has access to via an internal microservice.
for the company! But first, Cal needs to prove the
market exists—and they’re going to do it with the From there, the agent shifts from market research
company’s AI agent. to planning. Using machine learning, it creates a
predictive function for the company’s pizza chain
Cal tells the agent to find EV charging locations locations near select EV stations. It forecasts
across the country and map how close they are to potential pizza demand, so stores can deliver the
food. The agent identifies a map database, writes perfect number of pizzas and customers can get
queries to the map API, and correlates the data to the pies fresh without long waits.
create a list of hundreds of EV charging stations
with food beyond walking distance. It’s a start, but In just one day, Cal and the agent took an idea
Cal needs more. through market research and planning to be near-
ready for implementation. They quickly draft a
They ask the agent to estimate potential demand proposal and send it to their lead for review. They
using the stations’ occupancy data as well as the can’t wait for tomorrow—maybe they’ll build
pizza company’s own sales data from local stores. something else brand new.
The agent responds that some EV networks provide
Technology Vision 2025 | AI: A Declaration of Autonomy 22

02

Your Face,
in the Future
Differentiating when
every interface looks
the same

What’s your AI’s personality? Enterprises are

imbuing autonomy across customer interactions,
but as they do, it is critical that fine-tuning for
experience is not overlooked. Personal touch,
unique voice, and individualization are key ways
companies build loyalty and trust with customers—
and right now it feels like that is everything AI lacks.
Personified AI can breathe life into experiences
and unlock customer relationships like we’ve
never seen before. As enterprises start to build AI
cognitive digital brains, they can codify their values
and personas into it, powering agents with the full
knowledge of the business and allowing leaders to
dodge the risk of monotony and instead build the
future face of their enterprise.
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 23

1877 1963 1971 2000 2005 2014 2016

Quaker Mills’ Quaker Ronald McDonald As part of a DARPA Salesforce.com launches IKEA unveils one of Amazon releases Lil Miquela debuts on
man is one of the first makes his first TV project, Carnegie Mellon its website, allowing the first customer-facing Alexa, an AI assistant Instagram, the first AI
corporate mascots48 appearance and University creates online access to its CRM chatbots52 with a persona53 influencer to go viral54
becomes an icon49 Harpy, foundational to systems51
modern Interactive Voice
Response (IVR) systems50

2024 2024 Future 2025 2026 2027 2028

Instagram tests Creator.ai OpenAI releases Real-time generative A celebrity influencer A major airline introduces Companies routinely
for influencers to connect GPT-4o with Advanced AI voice models become signs a landmark contract an AI agent that will use customer
with fans via chatbots that Voice mode’s human-like common in customer letting a company use autonomously generate conversations with
mimic their personalities56 speech56 service their personality not only personalized trip pricing AI agents as a data
in advertisements but and travel recommendations source to build
product lines, customer based on customer loyalty customer profiles
service, and digital programs, company offers,
experiences and affiliated partnerships

2030

With the rise of agents, machine

customers drive more than 20%
of many businesses’ revenue,
transforming sales strategies
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 24

The Big
insight, perhaps inconsequential at first glance. But
enterprises are creating the customer interface

Picture
that will define their next decade—and how they
design it has profound, far-reaching implications.

Businesses are teetering on the edge of a customer

engagement crisis. It’s one that started growing
during the digital era. When new digital channels
All algorithm and no soul makes the started emerging, and customer touchpoints
enterprise a dull experience. shifted to platforms and search engines, businesses
could connect to more customers than ever, but
Dull for customers, that is. Across industries, direct conversation with them became muted,
businesses are finding ways to implement and differentiation became more challenging.
generative AI. They’re looking for operational
efficiencies, ways to automate tasks, and ways Now, in the AI era, this engagement crisis is at
to scale impact. Many see customer-focused risk of being compounded. AI models and the
roles as a natural fit. They’re reinventing the growing autonomy of digital systems can generate
face of the business, using generative AI for massive opportunity, the chance to have one-on-
customer service or ad material generation and one conversations at unprecedented scale. But
integrating chatbot interfaces into products. if these models are fine-tuned for function, and
not experience, they’ll sound generic or bland
But if they’re not careful, every business is and leave all that potential unrealized. This won’t
about to wind up with the same face. just be a problem on the enterprises’ channels
either. Looking towards the future, it becomes even
Effectively, enterprises are imbuing the customer sharper as generative AI-based chat platforms
experience with autonomy. Agents, bots, and position themselves as the primary window to the
technology systems can drive sales, solve digital world. When a third-party agent invites your
problems, set meetings and do much more for companies’ agents into a conversation with
customers. But the foundation models driving many a customer, what will it sound like?
of these experiences are built by the same few How will you stand out?
organizations and are designed—intentionally—to
be as neutral-sounding as possible. It’s a simple
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 25

What makes the world interesting is diversity and Enterprise leaders know people are capable of reps or sales associates, most businesses already
personality in individuals, businesses, products, building powerful—and valuable—attachments to train their people to do it, so why not AI too?
and experiences. While leaders hunt for the value brands with strong personalities. These personas
machine-like consistency can bring, they need to manifest as everything from mascots like Tony Emerging efforts show they can—to great effect.
ensure they don’t overshadow the very things that the Tiger and Coca-Cola’s visage of Santa Claus, In April 2024, Instagram began testing Creator.ai,
create enriching and personal human experiences. to partnerships with celebrities or influencers, which provides a program that will let influencers
AI systems by design will be more consistent than the to a distinct brand voice like a greeting at a connect with fans through chatbots that mimic
usual human interaction and engagement, and that retail store or a certain style for advertisements. their voices.64 The bots will automatically message
can be a great benefit, but leaders must take extra They’re ways to signal values and characteristics fans on the influencers’ behalf, letting them
care to inject the enterprise's personality in them—or that humanize a brand, and have been used to communicate and build rapport with far more

There is a unique they are at risk of losing what makes them unique.

Peel back the layers here, and it’s easy to see

drive familiarity, loyalty, or joy in customers.

At the same time, people are beginning to

people than they could ever do by hand. These
are chatbots creating real-feeling connections
at a scale and to a degree of personalization

opportunity right that what starts as “the chatbot sounds generic”

portends a potentially existential challenge for
experience emotional, real connections with AI.
Nomi AI is creating AI companions that people
that has never been possible before.

now for enterprises

a company’s brand. can text in group chats or one-on-one.61 They have And SiriusXM worked with Sierra, a conversational
personalities and backstories; they even remember AI platform, to build Harmony—an AI that actually
This is not a call to turn away from autonomy or past conversations. In August 2024, Soul Machines, understands customers on their terms.65 Many

to leapfrog the AI, nor have the consequences of generic-sounding

experiences come into play yet. In fact, we’ve
a company that creates autonomous AI avatars,
released a survey finding Gen Z young adults are
brands often have a shorthand or unique way of
using words to talk about products, something

competition...
seen this generation of chatbots boost sales, generate starting to prefer support from AI assistants over that gets lost in translation when using “off
high quality leads, and solve customer problems.57,58,59 traditional self-help resources in areas like learning the shelf” AI. For instance, a typical AI model
Gartner® predicts that “by 2027, chatbots will become languages, boosting confidence, and exploring might not understand a customer who needs
the primary customer service channel for roughly financial strategies—potentially demonstrating the help “rejiggering the radio” or who asks where
a quarter of organizations and the primary lead appeal of talking to an agent at times when we they can find a particular band. But, trained
qualification tool for 10% of B2B sales teams.”60 But want human empathy, but not judgement.62 And on the right data and with the right knowledge
these are short-term, early-mover gains. Will they Salesforce CEO Marc Benioff said, semi-seriously, of the brand, Harmony might understand that
remain when every company has chatbots? There he will occasionally use ChatGPT as a therapist.63 a customer needs a refresh signal sent to the
is a unique opportunity right now for enterprises to car, or is trying to find the right channel.
leapfrog the competition, dodge the risk of neutral What we know is this: Businesses have long
and monotonous customer experiences, and depended on the power of brand personality. And What we have here is no ordinary brand tool.
update or course-correct brand strategies faster now people and AI are forming similar bonds Companies that successfully introduce AI with
and more easily than ever before. But to do so, as well. But despite the fact that businesses are personalities are looking at a technology that can
they will have to start thinking about autonomous leaning on autonomous systems to rapidly scale build relationships with customers and meet their
systems as more than just the function they are their customer experiences to new heights, we’ve needs at an unprecedented scale. They’re taking
performing. They must find a way to marry AI’s scale yet to see them training those systems to espouse personified brand, and personified AI, and they’re
and efficiency with the uniqueness of their brands. the personality and brand values they work so inventing the personified business.
hard to develop. Whether it’s customer service
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face,
Your ininthe
Face, theFuture
Future 26

Imagine this: when a customer interacts with models. It will require training from employees,
a business, they’re guided by a familiar face. A management, and even customers to accurately
chatbot that acts like a beloved mascot or an capture the values and presence the company
influencer’s persona the company has licensed. This wants to project. Ultimately, it’s likely entire teams in
agent shows up on the company’s channels or can organizations will be dedicated to this effort alone.
be invited into a conversation on other generative
AI-based platforms. Over time, it supports and gets It's time to bring generative AI to
to know the customer—as an individual beyond customer-facing roles with the
purchase history or demographics. And it builds a
trusted relationship, not just with its personality, but gravity it requires. Done poorly,
by taking relevant actions. It wields the power of the businesses could see the magic of
digital world, accessing data on products and deals their brands eclipsed. Done right,
from across the organization, and tapping into a
suite of technologies like data analytics and digital it’s the beginning of a new era of
twins, to meet the customer on their terms. And it’s customer relationships and trust.
doing all this autonomously and at scale—running
thousands of conversations at once. This is a level
of personalization, once only available to individuals
in extremely limited numbers, that AI models
trained on your company’s persona can bring to
every single interaction you have—with customers,
partners, employees, and even other agents.

Companies today are starting to address

dimensions of this challenge. For instance,
Walmart’s new mobile shopping assistant will
summarize product descriptions, compare
products, and provide customer reviews.66 NVIDIA
released a Retrieval Augmented Generation (RAG)
framework to be used with chatbots, allowing
interfaces real-time access to information across
a company.67 And platform startup Native AI uses
generative AI and digital twins to let companies
chat with digital twins of their customers to conduct
market research.68 But even with the technology
available today, few companies are truly capturing
the full scope of work that will be needed to
instill the enterprise’s personality into a host of AI
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 27

The Technology

Enterprises are clearly eager to bring generative

AI and greater autonomy to customer-facing roles.
But to do this in a way that will strengthen—rather
than harm—ties with customers, they will need a
deep understanding of the technological advances
letting chatbots move beyond generic-ness.

• Personified AI

• Multimodal Models

• Precision Data & Context

Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 28

Personified AI
What is it? Who is doing it today? Growing interest in AI for customer solutions
Since the release of ChatGPT, executives have become more outspoken about AI
Organizations are exploring how to take powerful AI leaders are starting to build personified AI involvement in customer interactions. As companies look to implement generative AI
generative AI models and give them human-like capabilities. OpenAI described its GPT-4o, for technologies, they need to carefully consider how the development of their
personalities. Through training, fine-tuning, and instance, as a step toward more natural interactions customer-facing AIs will shape how customers perceive their organization.
other techniques, they are starting to specify the between human and computer.69 It can talk
words or phrases a model uses, how it structures to people with near-human response time in a Number of mentions relating to AI and customer solutions
sentences, or its speech patterns and tone. And convincingly human voice. Previously, people could in earnings call transcripts, 2020 - 2024*
when chatbots are accompanied by visual avatars, use “Voice Mode” to converse with ChatGPT, but
facial expressions can also help AI feel real. these interactions converted spoken queries into 300
text, generated a response, then converted that
response back to speech—which took time. GPT- Executive interest in
How does it connect to the trend? 4o is a single model trained across text, vision, and 250 AI-driven customer 4x
audio, meaning all inputs and outputs are processed solutions surged
Personified AI might just seem like an interesting by the same neural network. after the release of
200 ChatGPT (Nov 2022)
exercise in how models present themselves, but it
will be a core differentiator for businesses in the
future—both in how customers will interact with the
150
business on their channels and how businesses will
be represented by agents in other AI applications.
100

50

0
2020 2021 2022 2023 2024*

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on earnings call transcripts of G3K companies; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 29

Multimodal AI models
What is it? Who is doing it today? Research into multimodal models has increased significantly
There is significant research interest in multimodal foundation models that go beyond
Companies are rapidly expanding the range of Runway’s Gen-3 Alpha is an AI model that can language to image, audio, video, and more. Creating models that can talk, see the world,
modalities foundation models can accept as produce realistic videos from simple text prompts and inhabit real-time embodiments is set to transform how people interact with AI and
inputs and produce as outputs. Image input and or still images.70 In July 2024, when the image-to- thus organizations.
generation is now table-stakes, and audio is quickly video feature was introduced, early demonstrations
becoming standard as well. The commercialization showed impressive quality and fast turnaround Number of research papers relating to Multimodal Foundation Models, 2020 - 2024*
of models that accept video inputs is also starting. times (for some, it took less than a minute to
And while video output isn't widely commercialized generate a video from an image). Runway is one
yet, it has been demonstrated in research and is of several companies building video generation
becoming publicly available. models. Others include OpenAI, Pika Labs, and 2020 50
Luma Labs.71

How does it connect to the trend?

19x
2021 147 DALL-E release
While text-based chatbots are becoming (Jan 2021)
increasingly common for simple enterprise
customer service, the addition of audio, image, and
video input/output will vastly expand the customer- 2022 383
facing ways AI can be used. These multimodal
models will be able to bring chatbots to life, letting
them connect with people more deeply and even
bring them to life visually with facial expressions 2023 1,842
and an environment that all reflect the brand.

2024* 2,834

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on ArXiv papers; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 30

Customer context is key

Executives are starting to understand the importance of customer context to personified
Precision data and AI. For personified AI to create meaningful customer interactions, it needs new types
of customer context. Demographic information and purchase history are not enough—
customer context the best AI will also leverage past conversations to inform future ones. And consumers
are ready for this new technology, with only 13% of consumers overall having negative
feelings about the use of generative AI for marketing and advertising.

What is it? Who is doing it today?

Precision data is key to businesses’ digital customer Microsoft Dynamics 365 is a platform that uses AI 76% of executives agree that
strategies today. It fuels the ability to deliver more to help companies turn their customer data and the conversational interactions using
relevant deals, contextualized offerings, and more. data across their organization into customizable, gen AI will become a way to gather 28%

Right now, though, businesses are stuck between generative AI experiences.72 According to Microsoft, relevant customer context.
customers that demand more relevance, and a this platform can leverage real-time insights about Agree Agree
regulatory and social environment that is limiting people to better personalize customer journeys and Strongly agree
Strongly agree
the ways they typically gather that data. to help sellers close more deals.

How does it connect to the trend? 48%

Source: Accenture Technology Vision 2025 Executive Survey, N=4,021

Personified AI can be a major reinvention of how
businesses gather useful customer data safely and
effectively. Through interactions with AI, businesses
have the opportunity to get to know customers
Overall, how do you feel about the use of Gen AI for
beyond purchase history and demographics— Consumers who are familiar with marketing
Overall, how and advertising?
do you feel about the use of Gen AI for marketing and advertising?
useful data, like their interests, needs, or even gen AI feel more positive (74%) about
conversation style can be gleaned from every the use of gen AI for marketing and
10% Very positive
interaction. What’s more is AI itself can be used advertising than consumers who are Very positive
Somewhat positive
less familiar with gen AI (42%). 35% Somewhat positive
to get consent in the moment, by asking the user 32%
Neutral
Neutral
for access to data that can create more relevant Somewhat negative
Somewhat negative
+32 pp Very negative
outcomes—access that can be removed and data 39% Very negative
37%
forgotten upon the termination of the conversation.
17%
12%
6%
8% 3%

Consumers less familiar Consumers familiar with

with gen AI gen AI

Source: Accenture Consumer Paradoxes Survey, Oct 2024, N=12,215. Consumers

less familiar with gen AI, N=4,389; Consumers familiar with gen AI, N=7,826
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 31

The Implications

...every To drive the most opportunity, bots need to get

personal with people. It’s no small task given
people’s general concern about sharing data
systems and the personified business needs
to be in lockstep with customers—directed
first by where they trust and benefit from this
the enterprise-level, explainability controls need to
be in place so human oversight can ensure lack of
bias and that the machine is operating as intended.

interaction and, among some, skepticism around AI. But the

potential of personified business is too big to miss—
technology, only progressing as their trust grows.
Building a personified business on these three

with AI is not
it could launch a new era of autonomy in customer The next area, and perhaps the most critical, is pillars of trust is non-negotiable. Autonomous
relationships and reshape entire value chains. So, data protection and privacy. It consistently ranks as systems and AI personification have incredible
handling the question of trust upfront is paramount. one of the chief concerns people and enterprises potential to transform customer relationships,

just building There are three key areas of trust that enterprises
alike have with AI. Fortunately, some solutions
to achieve the personalization AI promises,
building one-on-one rapport and loyalty, reshaping
how customers’ needs are met, and giving

or breaking
can address from the beginning: awareness and without compromising security and privacy, are customers the kind of individualized attention
education around the benefits of AI; the primacy already at play. Synthetic data can be an option that digital businesses have never been able to
of data protection and privacy; and implementing to keep PII away from models. Where private data deliver at scale. But all of it depends on trust. To

customers’ controls to understand autonomous decision making. is required, some techniques include ensuring
conversations are deleted or that functions are
get to know people well enough to meet their
needs and take productive, relevant actions
Starting with awareness might seem strange. Most restricted to remain on devices. For instance, to on their behalf, people need to be willing to
trust with the business leaders are used to leveraging technology
already in people’s hands, not needing to advocate
preserve customer privacy in its copilot Recall
feature, Microsoft makes sure all screenshots
open up. What good is a personified agent if
customers are too afraid to talk to it? Or spend

enterprise, for it. But generative AI is putting enterprises in

a unique—slightly awkward—position. Consider
are stored and processed locally.74 There’s no
one-size-fits-all, but security must be top of
the whole conversation asking for a human?

this: the Michigan State University Federal Credit mind as businesses introduce greater autonomy The personified business reinvention needs
but with the Union (MSUFCU) developed a chatbot named
Fran, and in a survey of some of MSUFCU’s low-
to customer-facing features. Or those very
interactions could be the enterprise’s greatest risk.
to start today—and it must be rooted in trust.
If trust in the technology can be nurtured,

technology itself. and medium-income families, 44% said they felt

nervous about using AI technology, but of that
percentage, 70% said they liked using Fran.73
Finally, the ability to explain an AI’s decision-
making will be deeply important to earning
whole workflows and value chains can be
transformed. Remember—it starts with brand,
but there’s no reason it should stop there.
trust. If a customer wants to know why they are
Businesses know people generally feel positive receiving a particular recommendation, they’ll
about the results of AI—but they’re also contending need to be able to probe the chatbot. What’s
with vaguely skeptical sentiment around the more, hallucinations continue to be a part of
technology as a whole. It means every interaction the generative AI experience, so if a user is not
with AI is not just building or breaking customers’ achieving an expected outcome, explainability
trust with the enterprise, but with the technology can help them identify where mistakes have been
itself. And it means the rollout of autonomous made—and escalate to a human if needed. And at
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 32

What's Next
It’s time to bring brand and AI personification together and launch a new era of personified
business. If enterprises can “hardcode” the feeling they want customers to have about
them into AI agents that span the business, they’ll be able to build trusting relationships
with customers at scales that have simply never been reached before.

If you’re an early adopter? If you’re just preparing to start? If you want to take a slower approach? How to preserve trust?

Infuse personality into your generative AI efforts: Develop a personification rollout strategy for Audit chatbots across your organization: Take care to keep your chatbots aligned:
Early adopters may already have chatbots, but the high-impact customer experiences: You might be further along than you think. Chatbots When it comes to brand, you don’t want chatbots
promise of personified business comes when chatbots Be wary of overconfidence. AI and chatbot novelty is bring a host of advantages, so it is no surprise taking on a life of their own, promising solutions
connect more effectively with people. This will require strong today, but as it starts to fade, personification many organizations have deployed them already. that don’t exist or giving your company a flavor you
input from business leaders and technologists alike. will become increasingly differentiating. Now is the That said, there might not be a consistent strategy. don’t control. Do a thorough review of the data you
Appoint a team of cross-functional leaders to define the time to identify high-impact brand touchpoints where Run a ground-zero audit of chatbots at your plan to train your AI agents on, both for the initial
strategy for personified AI. One of their first initiatives personality could soon be essential—maybe there organization and what your goals are for them. Do deployment and continuously as their features evolve.
should be taking inventory of the types of data that may are stagnant moments in the customer journey, or they provide value in unexpected ways? How are Work with AI experts to ensure the training data
contribute to this effort, and determining what your areas where intermediaries dilute your brand. Create they enhancing customer touchpoints? What’s the you use is appropriate for your goals. Additionally,
chatbot can and should access. Then create a technical a pilot program to test and refine how you might use gap between where you want to be and where you work with experts to set rules and boundaries for
strategy to bring personified AI to life with that data. personified AI at these touchpoints, tracking its value- are? Study competitors to gauge where you might your agents, limiting their areas of knowledge and
add before broadening out. be behind and the new areas ripe for deployment. vocabulary to better align with your brand.
Ensure guardrails around your autonomous systems:
Data security has long been critical to customer Build relationships in the growing personified Start the conversation between your Demonstrate restraint with data collection:
interactions, but with personified AI getting to know AI ecosystem: technologists and brand: People are starting to form closer relationships
customers even more deeply, that security will be Personified AI is still new with few vendors—though Defining your brand is not a new challenge. But with AI. But to form a personified brand people will
especially vital. Before you grant agents access to a growing number of companies are starting to programming that brand into a chatbot will be. How trust, businesses need to show restraint around
your organization’s information and permission to offer create personified AI products. Start seeking out the do you want personified AI to make customers feel? relationships they build and data they gather, as well
customer solutions, set detailed security rules with right partners for your own efforts, bearing in mind What should they sound like? Taking a slower approach as how they use that data. Prepare for regulatory
the technological infrastructure required to enforce that the models your organization uses today may helps you get this right. Gather input across your challenges around data privacy and security that
them. Define and continuously update AI explainability be incompatible with vendor solutions, or limit your organization. Prototype customer-chatbot experiences will vary depending on where your organization
policies to customers and within your organization. product options. Set an assessment criterion for with technologists and brand teams to get on the and customers sit. People may expect to customize
future partners that covers how they are approaching same page about how to represent your business their privacy settings or opt out of chatbots. You will
personification, whether they will work with your and what kind of relationship you want to build with need to design those controls from the start and be
current models, and whether your infrastructure will let personified AI. There could be quirks around your transparent with customers about what data you
you switch solutions easily to access new features. brand’s voice that you may not realize until you do are using and what they can and can’t opt out of.
this groundwork. Decide what data your model will
need to best embody your brand and stay current.
Technology Vision 2025 | AI: A Declaration of Autonomy Your Face, in the Future 33

A Portrait of the Future

Claire opens her computer, excited but nervous. The Instead, Claire glances at the third bank’s
input window for her generative AI agent blinks at message. “Hi Claire,” it reads. “My name
her. She begins to type, “Last night, my partner and is Bruno. I’m NorthRiver Bank’s first-time
I found a home that we’d like to buy. Now we need homebuyer agent. I know you’re probably feeling
to find a mortgage broker so we can make an offer.” overwhelmed to say the least. Going through
this process is a huge step, but I’ll try my hardest
“Exciting!” the agent responds. “I’ve pulled three to make sure you understand everything and
of the top brokers according to online ratings. feel comfortable every step of the way.”
Each has an agent I can bring into the chat.
Would you like to preview their introductions?” Claire smiles. She clicks on Bruno’s icon and
enters the chat with her agent. Over the next
“Sure,” Claire responds. Her agent shares the hour, Bruno and Claire discuss different loan
three messages with her, each from a different options, how they are structured, and what they
national bank. But she squints at her screen. mean for her finances in the long term. Claire
“I think you made a mistake,” she writes. “You connects Bruno to the real estate listing and
said you were going to find three. The first financial portfolio of her and her partner. Bruno
two messages are identical. See, they both helps Claire identify discounts and tax incentives
just say ‘Thank you for considering us, we offer where she lives as well as suggests how to save
great rates and look forward to working together. on closing costs, all before connecting Claire to
Please connect to our agent to proceed.’” a human broker to walk through final details.

“No, I’ve just queried them again,” responds

Claire’s agent. “That was the intended response
for each bank. Shall I ask them to regenerate?”
Technology Vision 2025 | AI: A Declaration of Autonomy 34

03

When LLMs
get their
Bodies
How foundation models
reinvent robotics
A watershed moment is underway in robotics as
foundation models transform robots from linearly-
programmed and single-purpose to versatile
machines that can reason. Using LLMs, VLMs, and
Robotics Foundation Models, leaders are building
a physical extension of the cognitive digital brains
that will power the enterprise. This is giving robots
ever-greater autonomy in the physical world—
allowing them to better understand physics and
their environment, have spatial awareness, interact
with people, and understand complex instructions
and take safe and accurate actions in response.
This transition will require the creation of a full, built
for purpose stack, and will grow the use cases and
operability of robotics while simultaneously making
them more flexible, re-purposeable, and enduring.
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 35

1950 1966 1973 2002 2006 2015 2016

Isaac Asimov’s I, Robot is Stanford Research Institute Waseda University creates iRobot launches Roomba, Willow Garage is founded The DARPA Robotics Boston Dynamics
published75 begins developing Shakey, WABOT-1, the first full-scale a floor vacuuming robot78 and begins work on ROS Challenge finals highlight releases a viral video
a robot that can navigate anthropomorphic robot that (Robot Operating System) many remaining barriers to of its humanoid
complex environments can walk and grip objects77 as the “Linux of robotics”79 general-purpose robots80 robot, Atlas81
and find and group boxes76

2022 2023 2024 Future 2026 2029 2030

Google Research unveils Stanford University, the Figure and OpenAI partner A few hospitals in A consumer electronics Humanoid and
RT-1, a transformer-based University of California to release Figure 01, a Japan and Denmark company introduces quadruped robot
model for robotics82 Berkeley, and Meta robot that reasons84 become the first to use a generalist robot production surpasses
introduce ALOHA, a generalist robots prototype for limited 100,000 per year
low-cost, open-source household tasks
robotics system83

2032 2035

Like app stores of the A logistics company

past, a market emerges deploys humanoid
to download new skills robots to address labor
for generalist robots shortages, delivering
packages to even the
most remote areas
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 36

The Big
Before, robots didn’t “think.” They did repetitive
tasks, rigidly. A small mistake or unexpected

Picture
occurrence could end a task in failure. A vague
command, “Can I have something to eat?” would
not compute. This inability to navigate or reason
through a world built for and lived in by people
is why robots have never been truly autonomous,
relegated to heavily controlled environments like
A silvery humanoid stands at a table with an apple, factories and warehouses and largely kept away
some dishes, and a drying rack. A man asks for from the general population. They were guardrailed
something to eat, and the robot hands him the to be with other machines and not mixed with
apple. When the man asks why, it explains that the human workflows or interactions. But now these
apple was the only edible item.85 limits are dissolving. Foundation models are
unlocking a new chapter as we begin to assemble
There is a seismic shift happening in robotics and a full stack of robotics hardware, software, and
AI—and this scene from early 2024 was one of AI models purpose-built for one thing: machine
the first major demonstrations. The robot, called autonomy in the human world.
Figure 01, had no human assistance. Using a large
vision language model (VLM) trained by OpenAI for Over the next decade, we will start to see robots
visual reasoning and language understanding, and casually and commonly interacting with people,
robotics startup Figure’s neural networks for fast, reasoning their way through unplanned tasks,
dexterous robotic action, it understood the person’s and independently taking actions in any kind
question, recognized the apple as food, and acted of environment. It is an opportunity for physical
on its own. reinvention in industries far beyond the ones typically
leveraging robotics. Robots will be able to move into
Machine intelligence is moving into the physical customer-heavy environments, work in unpredictable
world, and robots are starting to demonstrate settings, communicate with anyone and take on any
reason and autonomy. We’ve long had robots, number of tasks, without needing to be specifically
but this is something new: the dawn of machines trained for each. Advances and adoption are coming
that can perform and coexist in truly human fast. Recent years have seen a meteoric rise in new
environments. startups, investments, and innovation, each tackling
different dimensions of robotics.86
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 37

Businesses in every industry, from traditional This would mean after training on video tours tasks down into smaller subtasks, to help them
industrial fields to areas that have never considered of an office space, robots can maneuver and adjust to unexpected disruptions more flexibly.90
robots, need to start imagining what their business guide people around, and even answer context-
could become and achieve in a world where robots based questions like “Where can I put this?”
are accessible, flexible, and—for all intents and While each of these three dimensions alone would be
purposes—think for themselves. Essentially, a world Evolution #2 Communication
where OT can tap into the power of AI autonomy.
enough to advance robotics, combined they are spurring the
The second dimension goes beyond robots biggest step change in robotic history: the rise of generalists.
This starts with understanding exactly how understanding their own space, to how they
foundation models are transforming robots. They interact with others in that space. Machine
are driving rapid advances across three dimensions: communication is moving away from lines of code Every enterprise must take note as this is no mere
contextual understanding, communication, and towards natural language, which will allow anyone technology upgrade. Generalists will kickstart a
planning and action. to interact with these complex machines. Robots period of robotic innovation rich with opportunity
like Figure 01 or Google Research’s generalist for every industry that has physical operations—
Evolution #1 Contextual Understanding robotics model PaLM-E can be talked to in plain from retailers to healthcare, insurance, and more.
speech. In a 2023 demo, a person asked a mobile How businesses think about and design physical
When it comes to context—robots are getting better robot to bring them a bag of chips, and PaLM-E environments, processes, and workflows, and the
not just at recognizing but understanding things. arranged for the robot to open a drawer, take limits on what can be physically accomplished, are
Figure 01 saw the apple and knew enough about its out a bag, and hand it over.89 It used an LLM and all about to change.
weight, texture, and edibleness to hand it over. This sensor data from the robot to understand the
is because LLMs and VLMs are linking information spoken command, devise a plan, and execute. Look at how robotics hardware is evolving. Perhaps
from the internet to the physical world, helping the strongest signal of the generalist evolution,
connect the dots between what’s perceived and Evolution #3 Planning and Action humanoid robot development has boomed in
the contextual knowledge needed to intervene or the last 12 months, and they’re making their way
interact in the right way. Numerous organizations However, humans don’t always communicate into warehouses and factories.91 This exploration
are now building these systems or datasets precisely, and we don’t account for every of more multipurpose robot form factors is an
to support them—like Stanford University and possible variable. This brings us to the third acknowledgement that to realize the potential of
Robotics at Google, which together compiled the dimension: planning and action. Robots used to generalist brains, robots will need bodies to match.
PhysObjects dataset of hundreds of thousands of have very limited actions because each action To navigate a world built for humans, they will have
annotations on physical concepts and properties.87 needed to be meticulously programmed. But to move like humans too. And quickly this will go
today robotic systems can use LLMs to interpret from pilots to the norm: Goldman Sachs found the
Context around individual objects isn’t enough abstract prompts or break complex tasks global market for humanoid robots could reach $38
though—and researchers are also working at down to smaller steps, greatly expanding their billion by 2035.92
giving robots a better understanding of the space ability to understand and react to commands
they are in. For instance, a navigation technique or situations they haven’t specifically been The autonomy granted by the evolution of robotic
proposed in a 2024 paper combines the common- trained for. Engineers at MIT, for instance, are bodies and “brains” points to a future where robots
sense reasoning of a long-context VLM with a connecting robot motion data to the “common are widely capable and widely adaptable—an
navigation policy based on topological graphs.88 sense” of an LLM to let robots break household important shift that can change the economics for
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 38

robot deployments. Unlike in the past, these robots waiting 30 seconds for a machine to respond to
can handle changing environments, complex and a request will break the real-world immersion, so
unpredictable work, and can learn new capabilities understanding how robots will handle requests,
and be redeployed. This means companies can be not just with human-like capability but at human-
more experimental with them, trial running new like speed, will require decisions about where
use cases, A/B testing different customer-facing models are hosted and how robots are networked.
deployments, and revising as needed. Robots are Finally, as foundation models deepen into other
becoming an investment that can change and areas of science, like physics and chemistry,
appreciate over time, and with this freedom, we can adapting these models to give robots an even
expect to see an explosion of new use cases and deeper contextual understanding of our physical
ideas emerge. environments will be the next big step forward.

Just think about the sheer range of impact And the technology hurdles are only part of the
that generalist robots, and greater freedom to pathway to enterprise adoption. Trust will also be
experiment with them, could have across industries. critical to true real-world viability. People won’t
The value that robots brought to industrial settings just need to work with the machines around them,
for decades—streamlined processes, cost but trust and depend on them as well. Enterprises
reductions, 24/7 operations, protecting humans looking to spur their robotics revolution will have
from dangerous work, eliminating human error—all to consider new dimensions of the human-machine
of this is on the table for everyone now. Imagine relationship. Like, what happens when responsible
small businesses that never close. Warehouse AI practices migrate into the physical world? What
stores where robotic customer service agents decisions will a machine make when faced with
know your purchase history and help with heavy an emergency or impending accident? What will
lifting. Stadiums, airports, and shopping centers it mean to steward a new generation of physical
where security never tires and lines move quickly. (and possibly energy intensive) technology in a
Autonomous delivery systems that go right up to sustainability-conscious landscape? Enterprises
your door and ring the bell. Even home robots to need to be at the forefront of not just technology
help with cleaning, watching pets, and more. integration, but social integration of these machines
if they want to achieve the benefits.
Of course, as foundation models pave the way
to this new generation of robotics, enterprise Now is the time to start building your robotic future. As
leaders will also need to keep a keen eye on
other emerging technology innovations that will generalists learn to navigate new environments, connect with
be critical to making these machines real-world people in them, and “think” through problems autonomously,
viable. Extremely low power consumption will their reach and impact will vastly and rapidly expand. Robots
be a core competency to allow them to run with
the duration and efficiency needed to tackle are about to go places they’ve never gone before, and it’s
tasks they will be given. Latency will also be key: up to you to reimagine your business for this new world.
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 39

The Technology

Today’s skyrocketing potential for generalist robots

began with the emergence of foundation models.
They opened the doors to the possibility of real-world
reasoning, adaptability, and autonomy—and to bring
this to life, leaders are exploring new techniques,
working to resolve training data shortages, and
building a new generation of robot hardware.

• Foundation Models for Robotics / Embodied AI

• Robot Training / Data Collection

• Multipurpose Robot Bodies

Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 40

Foundation Models for Robotics / Embodied AI

What is it? Who is doing it today?
Embodied AI is AI specifically designed to interact Google Research has been a leader in this
with the real world through a physical body. The field for years. In 2022, Google’s Gato model Research into embodied AI is steadily increasing
technical advances driving the generative AI showed that robotic inputs and outputs could The development of foundation models is driving a renewed interest
revolution have sparked new interest in embodied be tokenized and adapted into a language in the concept of embodied AI. Continued research and advances in
AI, with some researchers believing it is the model.93 Since then, the company has remained embodied AI are poised to expand the capabilities of robots significantly.
key to achieving the next leap in AI capability. a pioneer, creating a series of progressively
Various approaches for embodied AI are being stronger robotics transformer (RT) models: RT-1, Number of research papers relating to Embodied AI, 2020 – 2024*
tested, including incorporating LLMs and VLMs RT-2, and RT-X. Each improves on the one before,
into robotic systems on top of robotic control aiming to create generalizable robot policies
policies, as well as creating end-to-end robotics that can be adapted to a range of robots.94,95,96 700
foundation models trained on robotics data.
600

How does it connect to the trend? 4x 605

500

Innovations in embodied AI are critical to

building robots that have greater contextual 400
understanding of the world, communicate 384
with people using natural language, and plan 300 Release
ways to complete complex actions (even of Gato
& RT-1
ones they haven’t been specifically trained
200
for) or adapt to changing environments.

75 141
100
30

0
2020 2021 2022 2023 2024*

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on ArXiv papers; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 41

Robot Training / Data Collection

What is it? Who is doing it today? Researchers are consolidating robotics efforts
Open X-Embodiment is an ambitious project to consolidate open
Part of the power of foundation models today is Many groups have started creating open robotics robotics datasets into a centralized collection for the training of
the robust training data used in their creation. But and embodiment training sets, hoping to accelerate generalist robot policies. The dataset consists of data from 21 institutions
where LLMs leveraged all written language across AI for robotics much like how ImageNet helped and 22 different robots completing 527 skills, highlighting the growing
the internet, robotics training sets have been accelerate computer vision advancement. One number of organizations participating in creating open robotics datasets
scarce. Now, that’s starting to change. New robotics example is the Open X-Embodiment Dataset, and a large increase in the volume of training data in recent years.
datasets, and the ability to leverage other data which is a compilation of 60 open robotic
types in combination with simulation and real-world datasets.99 As for ongoing data collection for
reinforcement learning approaches, are helping to training robots, one example comes from the Total Number of Datasets in Open Total Number of Robotic Episode Examples in
fill the gap. auto industry. Tesla sources training data from its X-Embodiment by Publication Date Open X-Embodiment by Publication Date
global fleet of cars to improve its Autopilot and
Full Self-Driving features, selectively triggering
How does it connect to the trend? to collect only the most relevant data.100 80 3.0M

12x
70
If companies want robots to work in different 2.5M

environments and accomplish a wide variety of 60 ~4x

tasks, their training data must reflect that variety. 2.0M
50
Assembling initial training sets will be important,
but equally important is establishing a framework 40 1.5M

for ongoing data collection to power subsequent

30
training and learning. Some companies are 1.0M

deploying robot fleets to gather their own real-world 20

training data.97 Others are leaning into synthetic 0.5M
10
data, especially for training in simulations where
robots can confront nearly limitless scenarios.98 0 0.0M

2020 2021 2022 2023 2024* 2020 2021 2022 2023 2024*

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on Open X-Embodiment Collaboration dataset; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 42

Multipurpose Robot Bodies

What is it? Who is doing it today?
In the past, many robots were built for a single Many companies—traditional robotics Investment in humanoid robotics takes a big step
function; their software and physical design were manufacturers and startups alike—are working on Despite a challenging investment market, funding and deals related to humanoid
both determined by the task they were created multipurpose robot bodies. Automotive companies robotics had a major increase in 2024. The versatility of their form makes them very
to do. But recently, more multipurpose form are showing especially strong interest. Some, like applicable for a wide variety of use cases.
factors have entered the market. The hardware Tesla, Hyundai (owner of Boston Dynamics), and
components inside these bodies are more purpose- Toyota, are developing their own robots.101,102,103 Total funding (USD) and number of deals relating to humanoid startups, 2020 – 2024*
built for robotics than ever, and the bodies they’re And they're starting to push past development into
composed into are increasingly general-purpose, deployment—a BMW plant is using an OpenAI-
Total Funding (USD) Number of Deals
meaning they can take on multiple types of tasks. powered self-correcting humanoid.104 And Magna, $1.4B
Humanoids are a prime example, but robotic arms, a car manufacturer, will be piloting a humanoid
44
dogs, even some drones are also getting more robot from Sanctuary AI in one of its facilities.105 $1.2B
versatile capabilities. $1.19B
35
$1.B

How does it connect to the trend?

$.8B
A generalist AI robotics model is critically limited
without a generalist robot to run it on. Humanoids 19 $.6B
are appealing because of their versatility. They will
fit quickly and easily into a human-centric world,
$.4B
accessing physical spaces and participating in 10 $0.47B
workflows initially designed for people. Videos 8 $0.36B
of humans doing various tasks can also become $.2B
training data for humanoids. $0.18B
$0.03B
$.B
2020 2021 2022 2023 2024*

Note: *2024 contains partial data through Oct 2024

Source: Accenture Research analysis on deals from CB Insights. Analysis includes deals with undisclosed
funding amounts and excludes the investment type “corporate majority”; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 43

The Implications

People’s The emergence of generalist robots will steadily

bring more AI autonomy into the physical world over
the next decade. It may sound like a lot of lead time—
machines to learn. But allowing data to be collected
could open liabilities that some enterprises might
not want to take on.
inspire—fundamentally different business models.
Whether an organization has a long history with
robots, or has never used them before, everyone

acceptance of but don’t get comfortable. This new generation of

robots will fuel a major reinvention of enterprises’ There are also going to be physical safety questions.
is starting from scratch here. Leaders need to
explore questions like how they will access these

these machines physical presences. Across all industries, they’ll be

able to redesign operations, power new services, and
engage people and the world differently. But much
There are overt issues that get a lot of attention: the
potential for a robot to be hacked by a bad actor
and used for malicious intent, or for a malfunction
machines, how they fit into cost structures, and how
the answers to these questions will impact the use
cases they prioritize or partners they engage.

will be critical like digital agents—perhaps even more so—these

machines will need a trusted environment to flourish
to result in physical harm. But there are also
questions about what happens when a machine is Consider this: generalists’ multipurpose capabilities

to enterprises’
in. Which means if enterprises want to maximize operating as intended. Think about a security robot mean companies can explore novel deployment
the opportunities, there’s a lot of preparation and that’s faced with a breach or violation, or a robot in strategies. In the past, robots were often huge
groundwork to cover now. the middle of a construction site emergency, where investments installed and programmed for just

success as they First, and perhaps most consequentially, is

injury may become unavoidable. These machines
may be put in a position to make choices about how
one task. If a company shut down an operation,
they could resell, but to a relatively limited market.

reimagine how
how enterprises will extend their responsible AI and where to minimize harm. Now, generalist robots’ adaptable bodies and ease
practices into the real world. At this point, many are at switching between tasks mean robotics-as-a-
familiar with responsible AI and have established The framework and governance of this decision- service models may become the most appealing

they operate and practices and policies, but robotics bring new and
challenging dimensions to those efforts.
making will be crucial for any enterprise. Leaning
on the established thinking and principles of
option to many. What’s more, being able to rent just
the machines you need means experiments can be
responsible AI will be a good start, but leaders started up with less hardware investment, and the
do business... Think about data privacy: there are already
standards like the right to be forgotten and
should be planning now for the unique situations
their environment poses. People’s acceptance
ability to redeploy or return them will make testing
new use cases less financially risky.
approval needs around data collection. But what of these machines will be critical to enterprises’
happens when a robot—which needs audio and success as they reimagine how they operate and This is just one new business model option—but
visual sensors to navigate—goes out into the do business in the human world and with humans now is the time to be thinking about them, weighing
world, constantly collecting information about its around them—and responsible practices is the very value against risks, and building your strategies.
surroundings. How will those machines intersect baseline needed for building that trust. With robotics-as-a-service, some risks might include
with laws around digital privacy or the right to film questions around data collection, what’s proprietary
in public or record without consent? If data needs Secondly, enterprises need to trust their investment. business knowledge, and what can be shared with
to be stored locally, it inhibits the ability for the This new generation of robots will require—or providers. And providers are starting to address
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 44

these tradeoffs too. Covariant, for instance, is a behind them, but in many cases, they’ll be adopted has even gone as far as agreeing to buy power
robotics-as-a-service startup that has worked to by businesses and locations that haven’t dealt with in the future from nuclear fusion startup Helion
demonstrate the value of widespread fleet data. this kind of asset before. Mismanaged, energy costs Energy.110
In 2017, the company set out to build the first could quickly escalate.
foundation model for a universal robotics picking
We may still be years away from
system—an extremely difficult task as the objects As companies plan their robotic transformations,
for picking can be very different shapes and sizes.106 there has never been a better time to explore generalist robots populating
It required a massive training set that didn’t yet smarter energy strategies. And there are a lot of storefronts, interacting with people,
exist, so the company built connected robots, now options. Load shifting lets users move energy
or going out on their own to assess
deployed to warehouses around the world, to share use from peak to off-peak times, for cheaper or
live data and learnings across the fleet to improve greener energy or to avoid over-taxing the grid. insurance claims or repair telecom
performance for everyone. Virtual power plants are networks of small-scale lines. But what’s clear is that there’s
energy sources and flexible loads—like electric
a lot to do before we get there. Any
And finally, businesses need to establish a trusted vehicles, home batteries, or robots—that can
environment for the long term, which leads us buffer a grid during energy disruptions. And company that sees the promise of
to sustainability. With ESG goals and customers’ microgrids are independent systems meant to these adaptable and intelligent robots,
increasing concerns, it’s a simple fact that no provide stable power to important areas. Each of
and wants to be part of the robot
physical reinvention can happen today without these techniques can help new robotic operations
sustainability at the forefront. run cleanly, consistently, and without over- future, needs to get to work.
burdening the grid. And some companies are
Think about the attention foundation models already already putting them to use. Google data centers
receive for their energy consumption. Where novel have been shifting some high-energy computing
technology developments were once associated tasks to when and where a grid has carbon-free
with greater efficiency, now they’re coming with energy available.107
greater demands than ever. And frankly, anything
new businesses do is being met by higher Another approach may be to invest in or help
sustainability scrutiny from the public. We see it develop cleaner energy sources. In early 2024,
with generative AI, and it will be true for emerging Amazon Web Services bought a fully nuclear-
technologies like robots, quantum computing, and powered data center.108 And the Three Mile Island
space too. With robots specifically, they’re not only nuclear plant will reopen, following a 20-year
a visible representation of the energy-hungry AI Microsoft deal to buy its power for AI. Microsoft
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 45

What's Next
How can you prepare for a world with robots that reason? Throw away old notions of where
robots can go and which industries can use them. It’s time to get creative and start imagining
how generalist robots might transform your business in the next decade, as well as ensure
your business can access and experiment with them when the time comes.

If you’re an early adopter? If you’re just preparing to start? If you want to take a slower approach? How to preserve trust?

Chart your path to scale: Partner with robotics leaders: Track the robotics landscape: Bring your responsible AI practices
For companies already testing generalist robots, Now is the time to build relationships with robotics The robotics ecosystem is changing faster today than in into the physical world:
moving successful experiments into full production will companies and researchers that may be important decades. Don’t let the fact that generalist robots aren’t There are valid reasons why people might be scared of
come with new hurdles. As you enter this next step, vendors and partners down the line. Today, many of yet commercially ready ease you into complacency. It is robots. Think of a physical security breach happening,
ensure you are asking questions like whether you need them will be looking for real-world test cases for their critical to monitor this space and industry competitors. and a security robot taking action with serious
stronger relationships with robotics manufacturers, technologies. See what these players offer—and what When the time comes, you will need to know these consequences. Who is programming its decision-
how you can monetize robotics data in new ways, what data and technological infrastructure you need for a machines’ capabilities, limits, vendors, and successful making? Should it take action at all? Questions like
your governance policies are on data gathered from partnership to flourish. Explore ways your people can test cases to hit the ground running. this don’t always have clear answers. But as a starting
the real world, how your security strategies need to be practice working with generalist robots. point, businesses will need to be transparent with
updated, and what continuous robot maintenance and Run a robotics ideation sprint: their robots’ roles, how they make decisions, and
training plans you need. Lean into co-innovation opportunities: One of the best things to do right now is figure out what people can do if mistakes happen. Robots acting
There’s still time to be a leader in this space. While what generalist robots will bring your business. It is in a safe, ethical manner is a complex dimension
Push experimentation to new areas: most organizations will not build their own generalist not easy. Robotic solutions have not been an option of responsible AI that few organizations are really
Don’t limit yourself to your first successful proof-of- robotics hardware or software from scratch, partnering before for many organizations. And this is a new prepared for. So, start thinking about it now.
concepts. The bounds on how and where generalist with organizations that do could inspire new industry- generation of robots, so you can’t necessarily draw
robots can be used are not yet determined. True specific solutions—with you at the vanguard. Start from the past. If you were building an entirely new Position robots as co-pilots:
pioneers will continue exploring novel use cases looking for these innovation partnerships and business, and these robots were at your fingertips, The most important factor in your robot rollout is
and publishing data sets to advance the industry. To opportunities that could position you as a solutions what would you do differently? Conduct a design your employees’ trust in these machines and in you.
get started, audit what robots are already doing at provider for others down the line. thinking workshop to identify all the workflows Assuage fears of robotic replacement by framing
your organization and where they could go next, and newly viable for automation and places this next robots as co-pilots designed to improve employees’
remember that while not every endeavor will work, generation of robots could impact. Go big. experiences. Remember one of these robots’
each one that does can bring major advantages. greatest features is how well they communicate—
build a trust strategy around that, with a feedback
system for people working with these robots to
guide them to become even better co-pilots.
Technology Vision 2025 | AI: A Declaration of Autonomy When LLMs get their Bodies 46

A Portrait of the Future

Sometimes, Andy thinks the robots are like children. shelving that’s about to be rolled out nationwide.
He teaches them basic skills and trains them for their Andy quickly pulls up one of the robot’s digital
futures, and then they leave and mostly don’t need twins and watches it repeatedly attempt and fail to
him, except when they come home for maintenance complete the task. There is indeed a problem.
or call for help.
First, Andy runs a series of tests to check for
Andy just took a new job as a robotic fleet hardware failures or software bugs. Then he checks
deployment specialist at AllTrades, a major robotics- the robot’s plan of action and queries the system
as-a-service vendor. His last job was managing in natural language over why it stopped the action
contracts for an industrial equipment dealer. Never early. He thinks he knows what’s happening. The
in a million years did he think he would go from new shelves are made of a highly reflective material,
seeing heavy machinery as anything but exactly and the robot is seeing something in them that’s
that. But here he is. He and his team work with tricking its vision software.
their flagship humanoid, Jack. And thanks to Jack’s
versatility and the domain-specific fine tuning they Using VR teleoperation, Andy assumes control of
provide for each project, they have fleets working in the Jack and uses it to place various items on the
retail and doing security at a football stadium. Later shelves. He records the visual data, then uploads
this year, two Jacks are slated to be sent to the moon it to AllTrades’ Jack simulator. Within minutes, the
to help set up and operate an unmanned lunar lab system has generated countless new variations of
for low gravity experiments. the data, and hundreds of virtual Jacks are practicing
placing items on reflective shelves. After a few hours,
An alert interrupts Andy’s musings. It looks like today a fine-tuned model is deployed and validated on the
is one of those days when the robots need his help. physical Jack. Success! Best of all, this new model
A big retail client has reported an issue with their will be pushed out to all of the client’s Jacks, before
Jack robots stocking items on a new type of display those new shelves hit more of their stores.
Technology Vision 2025 | AI: A Declaration of Autonomy 47

04

The New
Learning Loop
How people and AI are
defining a virtuous cycle
of learning, leading,
and creating
As enterprises see the growing capabilities of
AI, they may be tempted to approach it similarly
to automation technologies of the past. But
the uniqueness of generative AI lies in the fact
that it is a learning technology, becoming more
capable and useful the closer it is to people.
Conventional automation would yield one-
time benefits, and risk creating a disenchanted
workforce. But approached correctly, leaders
can ignite a virtuous cycle between people and
AI: where the more people use it, the more it will
improve, and the more people will want to use it.
The positive reinforcement between people and
AI will be necessary groundwork for developing AI
cognitive digital brains, and unlocking this cycle
will be the engine that powers diffusion and helps
the enterprise achieve its AI-driven aspirations.
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 48

1760 1800 1890 1913 1964 1980 1994

The Industrial Revolution 75% of the U.S. paid labor The U.S. census Henry Ford installs the IBM releases the IBM Steve Jobs likens the Netscape Navigator
starts a shift from an force is in agriculture112 incorporates tabulators, first moving assembly Magnetic Tape/Selectric personal computer to a is released and
agrarian and handicraft cutting calculation time line for cars115 Typewriter, transforming bicycle—machines that transforms how
economy to machine from seven to three desk work116 amplify people’s abilities117 people access
manufacturing111 years113,114 information118

2000 2020 2023 Future 2025 2027 2030

2% of the U.S. paid labor OpenAI unveils GPT-3, Microsoft launches 75% of knowledge A major company More than half of
force is in agriculture119 which has abilities like Microsoft 365 Copilot, workers have starts a program to Fortune 500 companies
writing code120 calling it “the most employer-provided reward employees for have autonomous
powerful productivity tool generative AI tools automating tasks supply chain
on the planet”121 management systems

2032 2035

Unions and construction Most industries see

companies reach an profit-per-employee double
agreement on using human- over the last decade
robot teams in Europe
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 49

The Big
In other words, the more people use it, the better
it gets, and then the more people want to use it.

Picture The good news? People are already showing a desire

to work with it. AI is diffusing across the workforce
at unprecedented speed, and for perhaps the first
time in history, the technology driving automation
can be used by anyone. And this is why what
There is no doubt that work will be profoundly happens next is so consequential for enterprises—
different a decade from now—radically because this pace of diffusion will grind to a halt if
reshaped by generative AI. But ask employers people are uncertain about what the future will hold.
and employees exactly how this transformation
happens, or what the results will be, and it quickly When people think about automation, it is often
becomes clear we are caught in a stare down. evaluated for how it can remove people from the
equation. But when it comes to generative AI, this
Employees, on one hand, certainly see the role would mean undercutting the very partnership
generative AI can play in their work. According to that amplifies its value. Using generative AI for
reports, 95% find value in it, and it makes many conventional automation would yield one-time
happier in their jobs.122,123 On the other hand, benefits while turning workers’ enthusiasm into
many workforce leaders are racing to capture disenchantment—and possibly halt all further
the immense advantages of AI, agents, and the progress. Rather, to ignite this virtuous cycle, and
increasingly autonomous systems they power— have AI learn and unlock its true value, it needs
and naturally, one of the first applications that to be used with people—not instead of them.
comes to mind is automation. It’s a situation
brewing uncertainty and distrust, and risks holding
back the technology’s adoption and potential.

But a solution exists; it lies in the nature of

generative AI. Generative AI is inherently a learning
technology. It can enhance and advance its skills
over time, ultimately improving its value to the
individual using it and to the organization as a whole.
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 50

This is why enterprises need to make a conscious expansive knowledge is trapped behind their commanding fleets of autonomous bots, with
choice to show their people the benefits, earn specific skillset or the parameters of their job, and limitless skills to do their bidding. But it’s equally
their trust, and be willing to approach this an Accenture Talent Pulse survey found that the important to make sure AI doesn’t overshadow
transformative technology in a brand-new way. number one barrier to acquiring skills is lack of time. workers and leave them feeling disconnected
By now, enterprises are no strangers to pushing or disengaged.
technology change across the organization. But By bringing people and AI closer
leading this change with your people is new Take Wayfair, where developers were equipped
territory. In the past, various technologies were together, leaders can transform that with Gemini’s “Code Assist.”125 It let them set up
pushed top down, and while there might have paradigm—today’s ever-growing suite environments 55% faster, with a 48% increase in
been delays to their full diffusion, enterprises were of tools can give people access to skills unit test coverage. And 60% of the developers
largely in the driver’s seat. This time, people need said Code Assist let them focus on more satisfying
to be the engine of that evolution—a challenge they don’t personally possess. work. Wayfair, like many other companies deploying
with the specter of automation looming over it. To coding copilots, successfully positioned AI as a
encourage full participation, enterprises can zero A marketer with a hunch about emerging trends can creator, unburdening a lot of initial work and letting
in on three distinct areas where this feedback access the data science skills to validate their ideas. employees “level up” to more of a systems designer
loop between people and machines is taking A truck driver who imagines an app to enhance role. But apply this model to another domain, like
hold, where enterprises should lean in, and where their inventory process can prototype it. And what’s writing, and you might be in trouble. Many writers
the advantages and benefits to people shine more is that as AI learns from how people deploy may find the initial creative process the most
brightest. Get it right, and leaders will unlock the it through these various tasks, it will be able to engaging part of work. Having agents write drafts
power of generative AI and foster a workforce broaden its skill set, becoming a better and better might let them “level up,” but it could leave them
that innovates from the ground up, rather than coworker with each use. By putting the full weight doing less fulfilling work at a much larger scale.
leaving people as bystanders to change. of the workforce behind any and every employee,
these tools can both create the autonomy workers Employees may be wary of the tradeoffs between
The first advantage that the worker-AI virtuous cycle need to take new initiatives, but also enhance utility and engagement. They might see the
offers employees is the ability to imbue everyone collaboration across different domains. Just look benefits of a skillful AI, but if it’s taking away
with “infinite” skills. This may sound exaggerated— at Toyota Research Institute, which developed a just the work they enjoy doing, it will leave them
but think about the creative ways people already generative AI assistant for designers that’s been less inclined to adopt it. To this end, enterprises
use generative AI in their personal lives. From tuned to engineering constraints.124 It aids the need to think about how to give people more
creating meal plans to finally getting thank-you creative process and understands things like drag “autonomy” in what AI they use. Every day there
cards written, they’re using it to do more, not less. and cabin dimensions, letting the designers tap are new tasks and automations where generative
It’s letting them meet goals, like sticking to a diet into engineering basics as they start on a new AI can be utilized—but not every use case is the
or keeping up with correspondence, overcoming project. It doesn’t replace designers or engineers, right one for your people. Businesses need to be
lack of time or skills to capitalize on their ambitions. rather it ensures a higher degree of viable designs, asking: What will these automations leave people
And this is an important insight for enterprises. Your improving their collaboration. doing? Giving people the autonomy to steer their
best innovators have always been those closest own AI use, and rewarding successful efforts and
to the development and delivery of work. They The second advantage for employees is ideas, will encourage them to imagine and test
know what the goals should be: where efficiencies navigating the pitfalls of low engagement. It’s ideas. And again, the learning loop will continue
lie and what customers value. But too often, this easy for enterprises to see the value in employees to allow people to concentrate on the aspects of
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 51

their job they like the most, while letting AI expand in chemistry to generate a list of potential drug
across the domains found to be most tedious. candidates, and from that, were able to identify 79
viable candidates and land on one to bring to trials.
This brings us to the third advantage: making people The whole process was designed around what work
part of the engine of change. was best for AI versus for people, and how the
two would collaborate. And going forward, more
People being more engaged in their work is a huge employees will want to work like this and will have
benefit, but there are major changes coming to ideas for how to do it in their fields and for their
the way we work and how enterprises organize. roles—becoming a powerful engine for innovation
Historically, when people feel that change is spread all across the business.
happening to them, versus being a part of it,
they are far slower to adopt the changes being Shifting from automation to putting autonomy in the
implemented. Enterprises already feel the urgency
to reinvent, but in many ways the speed at which workforce’s hands is about letting your people lead your
they do it comes down to their workforce. transformation—not just to streamline their work but
to do more and sow the seeds of future growth. Giving
When you give people the tools to expand their
skillset, to decide what is engaging them in their them the freedom to start working like this today,
work, it inherently positions them as a bigger part building small automations, finding efficiencies, and
of the broader transformation happening. It puts seeing which new innovations work and which don’t,
them in the driver’s seat of innovation, which will
only serve to boost the company’s generative AI will give you a jump start on the future, propelling
reinvention. Furthermore, it will actually bring the you far beyond what strict automation ever could.
reinforcement cycle to bear on trust between
people and AI. The more people have the autonomy
to use AI how and where they like, the more they
will trust using it. With more trust in the technology,
the more they will subsequently increase their use
and push the diffusion of the technology across the
enterprise.

When people aren’t resisting change, and rather

are empowered with AI, innovating at speed
becomes the new normal. At Insilico Medicine,
for instance, a series of fine-tuned models across
the drug discovery process are helping develop
a novel therapy for Idiopathic pulmonary fibrosis
(IPF).126 Scientists used a multimodal system trained
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 52

The Technology

Generative AI is the driving force behind today’s

workforce reinvention. It is enabling more general-
purpose, flexible, and smart automation techniques,
vastly expanding the type of tasks that can be
automated. And importantly, it’s democratizing these
techniques, transforming who can be empowered
to wield this autonomy to drive enterprise growth.

• Accessibility of Automation

• Agentic Workflows

• Physical Copilots
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 53

Accessibility of Automation Executives plan to make Gen AI accessible for automation

Gen AI is an incredibly powerful tool that gives individuals access to skills and the
ability to automate the organization. However, only 47% of executives say they
What is it? Who is doing it today? expect their organizations to make gen AI tools signficantly to fully accessible to
their employees to automate tasks and workflows over the next three years. While the
Generative AI is making automation more Camunda, a workflow automation provider, is plurality (49%) do say they will partially make them accessible, we believe this may
accessible. With LLMs and other foundation models, adding AI capabilities to its platform. One feature leave a lot of value unmet.
employees are accessing the insights and expertise called Copilot Suggestions lets employees describe
of AI with unprecedented ease. In coming years, a process in natural language, and then uses AI To what extent, if at all, do you expect your organization to make gen AI tools,
the advent of agents will exponentially grow this to automatically translate it into Business Process AI copilots, and other AI agents accessible to employees to automate tasks and
capability, as AI moves from insight to action Modeling Notation, the industry-standard graphical workflows in your organization over the next 3 years?
and more enterprise systems become imbued language.127 On the open source side, Semantic
with autonomy. And with open-source models Kernal is an open source software development
49%
also gaining momentum, soon more people and kit created by Microsoft, letting developers bridge
enterprises everywhere will be able to access the gap between LLMs and existing software
cutting edge intelligence. applications.128 Via plugins, LLMs can understand a 41%
request and write a call to execute an action within
a piece of software, letting agents directly operate
How does it connect to the trend? the application.

AI’s growing accessibility is what’s driving bottom-

up autonomy in the workforce. While we’ve had
no-code/low-code in the past, the adoption and
utilization of today’s natural language-driven AI
tools is growing far more rapidly and will touch far
more types of workforce tasks. People are already
bringing it to work—the question for enterprises
6%
is how to leverage these capabilities and workers’
4%
enthusiasm to reimagine their strategies. 0%

Not at all Minimally Partially Significantly Fully

Source: Accenture Technology Vision 2025 Executive Survey, N=4,021

Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 54

Agentic Workflows
What is it? Who is doing it today? Growing executive interest in agentic systems
AI agents have significant potential to transform businesses and
Agentic workflows are an approach to using agentic Kognitos, a generative AI automation company, executives are starting to recognize it. Over the past two years, executive
systems to solve complicated cognitive tasks. By is deploying AI-powered BPA solutions.129 Their mentions of AI agents and agentic systems has increased rapidly.
breaking problems into discrete steps, specific platform can automate complex workflows like
agentic systems can address each step, while an order management, contact centers, and financial Number of mentions relating to Agentic Systems in earnings call
overarching system orchestrates interactions. services, and let users set up these automations transcripts, 2020 – 2024*
An early version of this is AI-powered business using plain speech, enabling greater flexibility.
process automation (BPA). While BPA started as Meanwhile, Microsoft is developing AutoGen,
the automation of repetitive business processes, a multiagent framework that lets users define a 400
providers are integrating LLM solutions, shifting the set of agents, each with specialized roles.130 The
industry to more flexible and creative automations.
As agentic workflows become more sophisticated,
orchestration layer then determines how these
agents should interact with each other.
350 Executive
discussions around 41x
the agentic systems involved will become capable 300 agentic systems
of learning over time to improve the performance skyrocketed after
of their individual task and the output of the entire the release of
250
system. ChatGPT

200
How does it connect to the trend?
150
As agentic workflows take on a greater share of
task-specific work, they will become an accelerant 100
for your people. Organizations can use them to
create a layer of abstraction across technology, 50
handling lower-level tasks like writing code and
connecting pieces together. Instead of employees 0
asking “how can I write this software” or “what
2020 2021 2022 2023 2024*
software can perform this task,” they can ask “how
can an agent help me accomplish my goal?”
Note: *2024 contains partial data through Oct 2024
Source: Accenture Research analysis on earnings call transcripts of 13K+ companies; Jan 2020 – Oct 2024
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 55

Physical Copilots
What is it? Who is doing it today? Executives see opportunities for physical copilots
Physical copilots offer the potential to augment and automate how physical
Not all automations will be digital. Physical copilots Magna, whose manufacturing facilities assemble work gets done. Executives see performing dull repetitive tasks and
are about to bring a new level of autonomy to vehicles for leading automakers including assisting workers directly to complete a task (such as holding or lifting) as
the world. These are machines like robots or Mercedes-Benz, BMW, and Jaguar, has announced being the most impacted by physical copilots.
exoskeletons that can help people act or act on it will pilot the use of Sanctuary AI’s humanoid robot
their behalf. With generative AI, they’re beginning to “Phoenix.”131 The latest Phoenix robot, their 7th In which of the following ways, if any, would you expect workers in your
have greater contextual understanding of the world generation, is capable of automating new tasks in organization to leverage physical copilots over the next 5-10 years?
and the ability to take more flexible and general- less than 24 hours. An earlier version of the same
purpose actions in it. robot had been used in a Canadian retail store to
66%
complete activities that ranged from cleaning to
labeling, picking, and packing merchandise.132
How does it connect to the trend?
Embodied agents will allow robots to tackle a 45%
huge number of new tasks in the physical world
39%
and closely collaborate with people. Jobs that
have never been up for automation suddenly are— 31%
albeit with more of a focus on complementary
applications than full automation. Companies
should be looking beyond knowledge work when
considering what automation tools could empower 15%
their people to do more and drive growth and
innovation for the company.

Performing dull Assisting workers Allowing workers to Performing Performing dirty

repetitive tasks directly to perform tasks at a dangerous tasks tasks
complete a task distance

Source: Accenture Technology Vision 2025 Executive Survey, N=4,021

Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 56

The Implications
Many businesses In anticipation of workforce reinvention over the
next decade, enterprises need to carefully consider
Square Hospitality Group (USHG). He takes a unique
approach to his businesses, where he believes
2024 report, nearly two-thirds of employers felt job
candidates should have a foundational knowledge

know the values the long-term impact that diffusing AI across the
business will have on careers. Empowering workers
employees come first, and value for stakeholders
will derive from that.133 And there’s a lot to learn
of generative AI tools—but in the same survey,
over half of recent college graduates felt that their

they seek or the

with AI and setting them up to build their own from his hiring strategy. While acknowledging that programs did not prepare them adequately to use
automations may be rooted in a technology shift, technical skills are critical, USHG prioritizes a set of generative AI.134 While higher education worldwide
but people are the key to success. To guarantee a six characteristics called the “hospitality quotient”— is sure to respond to this gap, establishing

culture they want healthy workforce, enterprises must reevaluate the

talent they value and how they build careers and
intelligence, work ethic, empathy, self-awareness,
kindness and optimism, and integrity. They believe
programs and matriculating students take time.
So, for now, upskilling workers and preparing

to foster—but how
retain people. that new technical skills are far more easily taught new ones will land largely on the enterprise.
and acquired than these emotional traits.
We’ve long thought about employees through Finally, consistency of values and know-how with

many carry that the lens of skills. One person has mastered
Adobe InDesign, another knows Python, and
Now is the time for other organizations to shift how
they see and think about employees too. To start,
tooling will be nothing if workers don’t have the
time or incentive to innovate. Establishing clear
so on. Companies assess and deploy people the values and culture of a workforce are poised to governance models and ensuring there is room in
forward into their based on skills; they design workflows and
processes around them; and as people improve
become of much higher-order importance than the
specific skills a person is initially hired for, because
the workday to take on forward-looking projects
will be essential for employees to experiment with

talent pipeline? skills or gain new ones, they advance. So, what
happens now that we’re putting autonomy into
it is inevitable that those skills will shift and evolve
as AI capabilities continue to grow. Many businesses
automation and for innovation to flourish at every
level of the enterprise.
people’s hands and giving them near limitless know the values they seek or the culture they want
skills? We need to move past the “T” shaped to foster—but how many carry that forward into their In a world where skills are becoming more fluid,
model of deep skills development for career talent pipeline? Into where they hire from, how they and employee-driven autonomy is the backbone of
growth. We’re not just asking people to get recruit, and the performance they reward? innovation, workforces will need a new structure in
better at what they do anymore—the most place to guide their careers and ambitions. People
critical employees will be those that innovate Enterprise leaders will need a new focus on training need to understand what success in their roles
and fundamentally change what they do. Most too, which will inevitably become a consistent and looks like now, what their career journeys will hold,
businesses aren’t prepared for the disruption important part of jobs. While emerging AI tools and that the company is invested in them and their
that will bring to how they hire and develop are increasingly intuitive, the boundless potential futures. Only by building this trust with employees
talent. As skills become more fluid, talent that they hold will only be accessible if the people on will businesses be able to fully step into this next
espouses the values and culture of the organization the job know what tools are available and how to generation of talent and capture the full benefits of
will be those best equipped to succeed. use them. Right now, there’s a gap between the generative AI automation.
autonomous capabilities of AI and the people
One leader who has embraced this approach is who know how to maximize AI’s use. Bridging that
Danny Meyer, founder of Shake Shack and Union divide is a problem with growing urgency. In a
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 57

What's Next
Sparking the new learning loop between people and AI will be key to creating meaningful change and
continuing to drive the diffusion of this powerful technology across the organization. But for enterprises
to get it right, it will require building clear lines of communication with their people, implementing data-
driven strategies, and reimagining their talent strategy for a future where every employee has a sidekick.

If you’re an early adopter? If you’re just preparing to start? If you want to take a slower approach? How to preserve trust?

Build data driven strategies, and the tools Get specific with your automation strategy: Get on the same page: Prioritize employee buy-in:
to manage them: To start, there are a lot of AI tools out there—some you Even if you’re comfortable waiting, your employees To move fast and capitalize on the benefits of
Simulate and explore ways certain automations may might want your organization using and some that your may not be. You will need a clear policy about generative AI throughout the workforce, employee
impact individuals or groups. Additionally, as you adopt employees are already using. There are also decisions using AI at work to prevent uncertainties, even if buy-in is crucial. They are the ones who will wield
new AI tools and train employees to use them, work to make about what tasks can be fully automated, and your strategy is to hold off on integrating it into the AI tools and make this transformation successful
to identify the key skills that employees will need to where there is greater value in having your people the organization. Remember that the technology or not. But to harness employees’ best efforts, they
succeed in the AI age. involved and giving them room to innovate. To start is changing fast, and your employees may be need to trust that they aren’t being asked to automate
making these decisions, conduct agentic design ahead of you, so make sure you are consistently away their own jobs in the long-term. Stave off fears
Start an AI Bounty Program: interviews with employees and build a human-in-the- updating this policy. Additionally, establish the data, by bringing them into the planning. Be clear about
Reward your people for the best automations they loop automation solution roadmap. governance, and technological foundations needed your ground-up innovation goals and your long-
discover or build. Many companies know the value of to prepare for one day expanding AI use at work. term vision and solicit feedback before making
bug bounty programs. This tactic could be used to Understand what keeps workers engaged: sweeping changes to org structures and teams.
surface new ideas for automations, too. What’s more, The most valuable tool you have right now is the Track your industry:
the insights you gain from these kinds of programs knowledge your people have about their jobs: where The astounding capabilities of an AI-empowered Define and formalize new career growth pathways:
can help you build future employee role profiles or there are potential efficiencies, and also what they like workforce will happen faster than you think. Employee success and career progression are going
create upskilling initiatives to help align your current doing the most. As you build a skills transformation Vendors and solutions may emerge that make this to look different in a workforce with generative AI
workforce with your organization’s new goals. roadmap, learn from organizations with a track record kind of automation more approachable. Establish tools and near-infinite skills. Make sure you’ve laid
on distributed automation and talent retention. Open a team looking at the workforce impacts of AI in out guidelines for growth, so your employees know
conversations directly with employees to use their your industry. Set quarterly meetings to review what is expected of them and can set goals. Failure
knowledge to maximize engagement. findings so the enterprise isn’t caught off guard. to do so will leave employees feeling directionless.
Maintain a “watch list” of vendors and determine Shape an inspiring narrative for your employees’
your assessment criteria for future partners. future. Lead educational programming to build AI
competency across all levels of your organization.
Technology Vision 2025 | AI: A Declaration of Autonomy The New Learning Loop 58

A Portrait of the Future

Sam is creative director at BlueDot Apparel—a And one designer is trying something very different.
sustainable fashion brand—and her team is He requested a tabletop weaving machine and
gearing up to start next year’s product line. She image-to-CAD tool that can turn 2D images into real-
kicks off the season with a challenge: “Ask for world 3D prototypes in under an hour. He’s hoping
any AI tool you need; use them how you want. the chance to physically test more options will
But give me designs I’ve never seen before.” give him the freedom to be bolder in his designs.

Her designers jump in. They’re fresh from a Sam is thrilled to see the energy. She thinks
generative AI training program, where they they’ll deliver the company’s most innovative
spent two weeks learning fundamentals ideas yet. And she has her own AI experiment
and experimenting with design automation underway too. She’s working with a startup to
tools. They’re ready to test their ideas. create hundreds of virtual AI personas of their
customers, which will give feedback on their
Some use a mood board copilot to inspire their products before they’re actually released. Then,
creativity. The copilot follows what the designers’ with this feedback and her own expertise, she’ll
put on their boards, finds similar images, suggests evaluate her designers’ work, making final
color palettes, and even generates designs to selections with the likeliest chance of success.
help them get started. Others begin with hand
drawing. For them, imagining something new is
the best part. But they’re still using AI assistants,
tuned on material characteristics, to help pick
the right fabrics and construction techniques.
Technology Vision 2025 | AI: A Declaration of Autonomy Authors 59

Authors

Karthik Narain Accenture Research

Group Chief Executive of Technology and CTO

LinkedIn Prashant P. Shukla, PhD

Renee Byrnes
Abira Sathiyanathan
Donovan Griggs
Gerry Farkova
Technology Vision 2025 Lohith Kumar
Rohan Girish Amrute
Editorial & Research Team Mariusz Bidelski
Gargi Chakrabarty
Ari Bernstein Abhishek Gupta
Kevin Gallagher
Rebecca Kelliher Carrie Kleiner
Paul D. Johnson
Naomi Nishihara Ignacio Mamone
Anna Marszalik
David Strachan-Olson Abhishek Kumar Mishra
Sandra Najem
Vincenzo Palermo
Shruti Panda
Linda Ringnalda
Shruti Shalini
Joanna Szpadkiewicz
Ezequiel Pérez Vázquez
Technology Vision 2025 | AI: A Declaration of Autonomy About the Technology Vision 60

About the
Technology Vision

For 25 years, Accenture has Accenture Technology

and Accenture Research
Horizon scanning research and data science
analysis—leveraging semantic text analysis and
As a shortlist of themes emerges from the research
process, the Technology Vision team works to
developed the Technology Vision collaborate on the annual LLMs to extract insights from datasets including validate and refine the set of trends. The themes are
research process, which academic papers, investment trends and earnings weighed for their relevance to real-world business
report as a systematic review this year included:
call transcripts—to uncover key developments and challenges. The Technology Vision team seeks
across the enterprise landscape trends shaping the evolving technology landscape. ideas that transcend the well-known drivers of
technological change, concentrating instead on the
to identify emerging technology A crowdsourcing effort to gather opinions themes that will soon start to appear on the C-level
trends that will have the greatest from innovation experts and researchers across agendas of most enterprises.
the organization, in addition to interviews with
impact on companies, government technology luminaries, industry experts, and
agencies, and other organizations business leaders.

in the coming years. A global survey of 4,021 executives across 21

industries to understand their perspectives and
organizational priorities regarding emerging
This year the trends look three technologies. The survey was fielded from October
to seven years into the future, to December 2024 across 28 countries.

while remaining relevant across

industries and actionable for
businesses today.
Technology Vision 2025 | AI: A Declaration of Autonomy About the Technology Vision 61

The ongoing story

2025 The binary big bang Your face, in the future When LLMs get their bodies The new learning loop
Technology trends

For the last 25 years, the Technology Vision

has investigated the biggest technology trends
impacting enterprises. While a new report
is produced each year, our new trends are
intended to build on—not replace—trends of the
past. To that end, we include this graphic each A match made in AI Meet my agent The space we need Our bodies electronic
2024
year to share connections around the broader Technology trends
technology evolution we are witnessing, and
how themes like sustainability, science tech,
digital ownership, advanced computing, and
others continue to shape the world around us.

2023 Digital identity Your data, my data, our data Generalizing AI Our forever frontier
Technology trends

2022 WebMe Programmable world The unreal Computing the impossible

Technology trends

Science tech
Digital ownership
Sustainability 2021 Stack strategically Mirrored world I, technologist Anywhere, everywhere From me to we
The unreal Technology trends
Technology Vision 2025 | AI: A Declaration of Autonomy Executive Survey 62

Demographics— 28 Countries 21 Industries Company Size (USD Revenue)

Executive Survey Argentina

Australia
100
200
Financial Services
Banking 290
Less than $5 billion 44%

4,021 C-Level and Brazil 201 Capital Markets 137 $5 to $9.9 billion 20%
Director-Level executives Canada 202 Insurance 300
Chile 85 $10 to $19.9 billion 15%
China 200 Communications,
Colombia 80 Media and Technology $20 to $49.9 billion 13%
France 203 Communications and Media 202
Germany 191 High Tech 201 $50 billion or more 7%
India 190 Software and Platforms 201
Indonesia 80
Ireland 105 Health and Public Service
Italy 190 Health 200
Japan 204 Public Service 500
Malaysia 80
Mexico 85 Products
Netherlands 90 Aerospace and Defense 60
New Zealand 90 Automotive 102
Saudi Arabia 80 Biopharmaceuticals 80
Singapore 85 Consumer Goods and Services 290
South Africa 100 Industrial Goods and Equipment 100
Spain 100 MedTech 85
Sweden 90 Retail 301
Switzerland 92 Transport and Logistics 101
Thailand 80 Travel 100
United Arab Emirates 85
United Kingdom 210 Resources
United States 523 Chemicals 171
Energy 200
Natural Resources 200
Utilities 200
Technology Vision 2025 | AI: A Declaration of Autonomy References 63

References

1. Yao, D. (2022, May 10). 25 Years Ago Today: How Deep Blue vs. Kasparov 10. Yim, D., Khuntia, J., et al. (2024, March 20). Preliminary Evidence of the 20. Perry, T., and Voelcker, J. (1989, September 1). How the Graphical User 30. Wiggers, K. (2024, May 21). Microsoft wants to make Windows an AI op-
Changed AI Forever. AI Business: https://aibusiness.com/ml/25-years-ago- Use of Generative AI in Health Care Clinical Services: Systematic Narrative Interface Was Invented. IEEE Spectrum: https://spectrum.ieee.org/graphi- erating system, launches Copilot+ PCs. TechCrunch: https://techcrunch.
today-how-deep-blue-vs-kasparov-changed-ai-forever Review. NIH National Library of Medicine: https://www.ncbi.nlm.nih.gov/ cal-user-interface com/2024/05/21/microsoft-build-2024-windows-ai-operating-system-co-
pmc/articles/PMC10993141/ pilot-plus-pcs/

2. Altman, S. (2023, February 24). Planning for AGI and beyond. OpenAI.com: 21. Looking back at 10 years of Microsoft 365 making history. (2023,
https://openai.com/index/planning-for-agi-and-beyond/ 11. New Milestone in AI Drug Discovery: First Generative AI Drug Begins Phase January 11). Microsoft: https://www.microsoft.com/en-us/micro- 31. AutoGen. (n.d.). Microsoft: https://www.microsoft.com/en-us/research/
II Trials with Patients. (2023, July 1). Insilico.com: https://insilico.com/ soft-365-life-hacks/stories/looking-back-ten-years-microsoft-365 project/autogen/
blog/first_phase2

3. Bove, T. (2023, May 3). CEO of Google’s DeepMind says we could be ‘just a
few years’ from A.I. that has human-level intelligence. Fortune: https://for- 22. Conway, A. (2024, January 23). On this day 28 years ago, Java 1 was 32. Amazon Q Developer. (n.d.). AWS: https://aws.amazon.com/q/developer/
tune.com/2023/05/03/google-deepmind-ceo-agi-artificial-intelligence/ 12. Roark, C., Jamison, S., et al. (2024). The productivity payoff: Unlocking released. XDA: https://www.xda-developers.com/on-this-day-java-1-re-
competitiveness with generative AI. Accenture: https://www.accenture. leased/
com/content/dam/accenture/final/accenture-com/document-3/Accen-
ture-The-Productivity-Payoff.pdf#zoom=40
33. AI-assisted application development. (n.d.). Google Cloud: https://cloud.
4. 2024 Work Trend Index Annual Report from Microsoft and LinkedIn. google.com/products/gemini/code-assist
AI at Work is Here. Now Comes the Hard Part. Microsoft.com: https://www. 23. Colvin, G. (2022, November 30). How Amazon grew an awkward side proj-
microsoft.com/en-us/worklab/work-trend-index/ai-at-work-is-here-now- ect into AWS, a behemoth that’s now 4 times bigger than its original shop-
comes-the-hard-part 13. The Evolution of Labor Markets and AI. (n.d.). Conversion Capital: https:// ping business. Fortune: https://fortune.com/longform/amazon-web-ser-
conversioncapital.com/the-evolution-of-labor-markets-and-ai/ vices-ceo-adam-selipsky-cloud-computing/
34. Wu, S. (2024, March 12). Introducing Devin, the first AI software engineer.
Cognition.ai: https://www.cognition.ai/blog/introducing-devin

5. Sacolick, I. (2024, March 5). Generative AI copilots: What’s hype and where
to drive results. CIO: https://www.cio.com/article/1309604/generative-ai- 14. Edwards, B. (2024, August 14). Research AI model unexpectedly attempts 24. The App Store turns 10. (2018, July 5). Apple: https://www.apple.com/
copilots-whats-hype-and-where-to-drive-results.html to modify its own code to extend runtime. Ars Technica: https://arstechni- newsroom/2018/07/app-store-turns-10/
ca.com/information-technology/2024/08/research-ai-model-unexpected- 35. Jassy, A. (2024, August 22). X: https://x.com/ajassy/sta-
ly-modified-its-own-code-to-extend-runtime/ tus/1826608791741493281

6. Rooney, K. (2024, May 22). Amazon plans to give Alexa an AI over- 25. Vaswani, A., Shazeer, N., et al. (2017, June 12). Attention Is All You Need.
haul—and a monthly subscription price. CNBC: https://www.cnbc. arXiv.org: https://arxiv.org/abs/1706.03762
com/2024/05/22/amazon-plans-to-give-alexa-an-ai-overhaul-monthly- 15. 2024 Work Trend Index Annual Report from Microsoft and LinkedIn. 36. Bg2 Pod. (2024, October 13). Ep17. Welcome Jensen Huang | BG2 w/Bill
subscription-price.html AI at Work is Here. Now Comes the Hard Part. Gurley & Brad Gerstner. YouTube: https://www.youtube.com/watch?v=bUr-
CR4jQQg8

26. Schick, T., Dwivedi-Yu, J., et al. (2023, February 9). Toolformer: Language
Models Can Teach Themselves to Use Tools. arXiv.org: https://arxiv.org/
7. Mickle, T., Chen, B., et al. (2024, May 10). Apple Will Revamp Siri to Catch 16. Miscellany. (n.d.). Lapham’s Quarterly: https://www.laphamsquarterly.org/ abs/2302.04761
Up to Its Chatbot Competitors. The New York Times: https://www.nytimes. revolutions/miscellany/paul-krugmans-poor-prediction 37. 60 Minutes. (2023, October 9). “Godfather of AI” Geoffrey Hinton: The 60
com/2024/05/10/business/apple-siri-ai-chatgpt.html Minutes Interview. YouTube: https://www.youtube.com/watch?v=qrvK_KuIeJk

27. Goldman, S. (2024, September 5). Marc Benioff has declared a ‘hard pivot’
17. Ada Lovelace's Note G. (n.d.). Project Lovelace: https://projectlovelace. to autonomous AI agents. Will it be enough for Salesforce to thrive in the
8. Knight, W. (2024, July 11). Google DeepMind’s Chatbot-Powered Robot net/problems/ada-lovelaces-note-g/ generative AI era? Fortune: https://fortune.com/2024/09/05/salesforce- 38. Raising the bar on SWE-bench Verified with Claude 3.5 Sonnet. (2024, Octo-
is Part of a Bigger Revolution. Wired: https://www.wired.com/story/goo- ceo-marc-benioff-ai-agents-agentforce-dreamforce-gen-ai-era/ ber 30). Anthropic: https://www.anthropic.com/research/swe-bench-sonnet
gle-deepmind-ai-robot/

18. Celebrating Penn Engineering History: ENIAC. (n.d.). Penn Engineering:

https://www.seas.upenn.edu/about/history-heritage/eniac/ 28. Create with Adobe Firefly generative AI. (n.d.). Adobe: https://www.adobe. 39. SWE-bench. (n.d.): https://www.swebench.com/
9. Marks, N. (2024, May 14). CaLLM, Cool and Connected: Cerence Uses Gen- com/products/firefly.html
erative AI to Transform the In-Car Experience. Blogs.nvidia.com: https://
blogs.nvidia.com/blog/cerence-generative-ai-in-car-experience/

19. Vaughan-Nichols, S. (2019, September 5). COBOL turns 60: Why it will 40. Reinventing with a digital core. (2024, July 16). Accenture: https://www.
outlive us all. ZDNet: https://www.zdnet.com/article/cobol-turns-60-why- 29. Wiggers, K. (2024, September 16). Microsoft Copilot: Every- accenture.com/us-en/insights/technology/reinventing-digital-core
it-will-outlive-us-all/ thing you need to know about Microsoft’s AI. TechCrunch:
https://techcrunch.com/2024/09/16/microsoft-copilot-every-
thing-you-need-to-know-about-microsofts-ai/
Technology Vision 2025 | AI: A Declaration of Autonomy References 64

References

41. Salesforce Developers. (2024, March 14). Get Started with Einstein 51. This History of Salesforce. (n.d.). Salesforce: https://www.salesforce.com/ 61. Silberling, A. (2024, April 3). I have a group chat with three AI friends, 71. Morrison, R. (2024, October 4). 7 best OpenAI Sora alternatives for
Copilot Actions for Developers. YouTube: https://www.youtube.com/ news/stories/the-history-of-salesforce/ thanks to Nomi AI—they’re getting too smart. TechCrunch: https://tech- generating AI videos. Tom’s Guide: https://www.tomsguide.com/ai/ai-im-
watch?v=PQJlqU2LHPo crunch.com/2024/04/03/nomi-ai-group-chat-three-ai-friends/ age-video/openai-sora-looks-amazing-but-you-cant-use-it-yet-here-are-
the-7-best-alternatives

52. Mazdak Rezvani, C. (2016, November 6). Why your chatbot needs a vertical
42. Google. (2023, December 6). Personalized AI for you | Gemini. YouTube: focus. VentureBeat: https://venturebeat.com/business/why-your-chatbot- 62. Soul Machines’ Study Reveals That Gen Z – ‘The Transformer Generation’ – is
https://www.youtube.com/watch?v=v5tRc_5-8G4 needs-a-vertical-focus/ Navigating Life’s Challenges with AI Assistants. (2024, August 21). Business 72. The new era of AI-powered business. (n.d.). Microsoft.com: https://www.
Wire: https://www.businesswire.com/news/home/20240821813861/en/ microsoft.com/en-us/dynamics-365

43. Introducing AI SDK 3.0 with Generative UI support. (2024, March 1). 53. The Alexa Personality. (n.d.). Amazon Developer: https://developer.ama-
Vercel: https://vercel.com/blog/ai-sdk-3-generative-ui zon.com/en-US/alexa/branding/alexa-guidelines/communication-guide- 63. D’Onfro, J. (2024, September 17). Salesforce CEO Marc Benioff says he 73. Chatbot or Human? Research Shows Preference is Based on Task. (2024,
lines/brand-voice uses ChatGPT as a therapist. The San Francisco Standard: https://sfstan- May 14). Commonwealth: https://buildcommonwealth.org/blog/chatbot-
dard.com/2024/09/17/marc-benioff-jensen-huang-dreamforce/ or-human-research-shows-preference-is-based-on-task/

44. Greicius, A. (2016, November 2). When Computers Were Human.

NASA: https://www.nasa.gov/centers-and-facilities/jpl/when-comput- 54. Parsani, P. (2023, November 22). The AI Behind Virtual Influencer Lil Mique-
ers-were-human/ la. Cut the SaSS: https://www.cut-the-saas.com/ai/the-ai-behind-virtu- 64. Maheshwari, S., and Issac, M. Ready for a Chatbot version of your favorite 74. Bott, E. (2024, May 30). Is Microsoft Recall a ‘privacy nightmare’? 7 reasons
al-influencer-lil-miquela Instagram influencers? you can stop worrying about it. ZDNet: https://www.zdnet.com/article/is-mic-
rosoft-recall-a-privacy-nightmare-7-reasons-you-can-stop-worrying-about-it/

45. Narayana, R., Dadlani, S., and Das, R. (2022, August 29). How Much Time 55. Maheshwari, S., and Issac, M. (2024, April 15). Ready for a Chatbot version 65. How SiriusXM drives listener loyalty with Sierra. (2024, February 13).
and Energy Do We Waste Toggling Between Applications? Harvard Busi- of your favorite Instagram influencers? The New York Times: https://www. Sierra.ai: https://sierra.ai/customers/siriusxm 75. I, Robot, (n.d.). Goodreads: https://www.goodreads.com/book/
ness Review: https://hbr.org/2022/08/how-much-time-and-energy-do-we- nytimes.com/2024/04/15/technology/instagram-influencers-chatbots.html show/41804.I_Robot
waste-toggling-between-applications

66. Walmart is building a GenAI-powered shopping assistant. (2024, June 6).

56. O’Donnell, J. (2024, September 24). OpenAI released its advanced voice Walmart Global Tech: https://tech.walmart.com/content/walmart-global- 76. SRI International. (2020, April 2). 75 Years of Innovation: Shakey the Robot.
46. David, E. (2024, September 16). Slack now lets users add AI agents from mode to more people. Here’s how to get it. MIT Technology Review: tech/en_us/blog/post/walmart-is-building-a-genai-powered-shopping- Medium: https://medium.com/dish/75-years-of-innovation-shakey-the-ro-
Asana, Cohere, Adobe, Workday and more. VentureBeat: https://venturebeat. https://www.technologyreview.com/2024/09/24/1104422/openai-re- assistant.html bot-385af2311ec8
com/ai/slack-users-can-add-ai-agents-to-their-workflow-with-new-update/ leased-its-advanced-voice-mode-to-more-people-heres-how-to-get-it/

67. AI Chatbot With Retrieval-Augmented Generation. (n.d.). Nvidia: https:// 77. WABOT-1: The first humanoid robot. (n.d.). BBC: https://www.bbc.co.uk/
47. Adobe Express. (n.d.). Slack.com: https://slack.com/marketplace/A7P- 57. Marino, S. (2023, November 17). 35+ Chatbot Statistics You Need to Know www.nvidia.com/en-us/ai-data-science/ai-workflows/generative-ai-chat- programmes/w3ct5ymx
35MCT0-adobe-express for 2024. LocaliQ: https://localiq.com/blog/chatbot-statistics/ bot-with-rag/

78. iRobot. (n.d.): https://about.irobot.com/history

48. Little, B. (2024, March 14). Historic Food Mascots. History.com: https:// 58. Joshi, S. (2024, January 18). 60 Exciting Chatbot Statistics That Explore Its 68. Hampton, J. (2023, April 6). Native AI Raises $3.5M Seed for AI-pow-
www.history.com/news/historic-food-mascots Growth. G2.com: https://learn.g2.com/chatbot-statistics ered Consumer Research. Big Data Wire: https://www.datanami.
com/2023/04/06/native-ai-raises-3-5m-seed-for-ai-powered-consumer-
research/
79. Wyrobek, K. (2017, October 31). The Origin Story of ROS, the Linux of
Robotics. IEEE Spectrum: https://spectrum.ieee.org/the-origin-story-of-
49. Nowak, C. (2023, February 27). The Real Reason Ronald McDonald is the 59. Beckman, J. (2024, February 29). 120+ Chatbot Statistics for 2024. Techre- ros-the-linux-of-robotics
McDonald’s Mascot. Reader’s Digest: https://www.rd.com/article/why-ron- port: https://techreport.com/statistics/software-web/chatbot-statistics/
ald-mcdonald-mascot/ 69. Hello GPT-4o. (2024, May 13). OpenAI.com: https://openai.com/index/
hello-gpt-4o/

80. Knight, W. (2015, June 9). Why Robots and Humans Struggled with
60. Gamble, C. (2022, September 16). Improve B2B Buyer Engagement DARPA’s Challenge. MIT Technology Review: https://www.technologyre-
50. Moskvitch, K. (2017, February 15). The machines that learned to listen. With Live Sales Chat. Gartner: https://www.gartner.com/en/docu- view.com/2015/06/09/110342/why-robots-and-humans-struggled-with-
BBC: https://www.bbc.com/future/article/20170214-the-machines-that- ments/4018765 70. Franzen, C. (2024, July 29). You can now turn still images into AI videos darpas-challenge/
learned-to-listen with Runway Gen-3 Alpha. VentureBeat: https://venturebeat.com/ai/you-
can-now-turn-still-images-into-ai-videos-with-runway-gen-3-alpha/
Technology Vision 2025 | AI: A Declaration of Autonomy References 65

References

81. Thomson, I. (2016, February 24). Google human-like robot brushes off 91. Yao, M. (2024, August 13). Humanoid Robots on the Rise: Industry Ad- 102. Hyundai Robotics. (n.d.): https://www.hyundai-robotics.com/english/ 112. Ruggles, S. (2007). The Decline of Intergenerational Coresidence in the
beating by puny human – this is how Skynet starts. The Register: https:// vances, Key Players, and Adoption Timelines. Topbots.com: https://www. United States, 1850 to 2000. University of Minnesota: https://users.pop.
www.theregister.com/2016/02/24/boston_dynamics_robot_improve- topbots.com/humanoid-robots-overview-2024/ umn.edu/~ruggl001/Articles/ASR-2007.pdf
ments/

103. Toyota Research Institute. (n.d.): https://www.tri.global/our-work/robotics

92. The global market for humanoid robots could reach $28 billion by 2035. 113. The Hollerith Machine. (2024, August 14). United States Census Bureau:
82. Gopalakrishnan, K., and Rao, K. (2022, December 13). RT-1: Robotics Trans- (2024, February 27). Goldman Sachs: https://www.goldmansachs.com/ https://www.census.gov/about/history/bureau-history/census-innova-
former for real-world control at scale. Google Research: https://research. intelligence/pages/the-global-market-for-robots-could-reach-38-billion- tions/technology/hollerith-machine.html
google/blog/rt-1-robotics-transformer-for-real-world-control-at-scale/ by-2035.html 104. Westfall, C. (2024, July 7). OpenAI-Powered Humanoid Robot Fills Spot At
BMW Assembly Plant. Forbes: https://www.forbes.com/sites/chriswest-
fall/2024/07/07/openai-powered-humanoid-robot-fills-spot-at-bmw-as-
sembly-plant/
114. Hollerith 1890 Census machine. (n.d.). Computer History Museum:
83. Zhao, T., Kumar, V., et al. (2023, April 23). Learning Fine-Grained Biman- 93. Reed, S., Żołna, K., et al. (2022, May 12). A Generalist Agent. Google Deep- https://www.computerhistory.org/collections/catalog/102630714
ual Manipulation with Low-Cost Hardware. arXiv.org: https://arxiv.org/ Mind: https://deepmind.google/discover/blog/a-generalist-agent/
abs/2304.13705
105. Heater, B. (2024, April 11). European car manufacturer will pilot Sanctuary
AI’s humanoid robot. TechCrunch: https://techcrunch.com/2024/04/11/
european-car-manufacturer-will-pilot-sanctuary-ais-humanoid-robot/ 115. The Moving Assembly Line and the Five-Dollar Workday. (n.d.). Ford:
94. Gopalakrishnan, K., and Rao, K. RT-1: Robotics Transformer. https://corporate.ford.com/articles/history/moving-assembly-line.html
84. Franzen, C. (2024, March 13). OpenAI powers a robot that can hand people
food, pick up trash, put away dishes, and more. VentureBeat: https://ven-
turebeat.com/ai/openai-powers-a-robot-that-can-hand-people-food-pick-
up-trash-put-away-dishes-and-more/ 106. Chen, P. Robotics Foundation Models and the role of data.
95. Chebotar, Y., and Yu, T. (2023, July 28). RT-2: New model translates vision 116. IBM Magnetic Tape Selectric Typewriter IV. (n.d.). Rhode Island Computer
and language into action. Google DeepMind: https://deepmind.google/ Museum: https://www.ricomputermuseum.org/collections-gallery/small-
discover/blog/rt-2-new-model-translates-vision-and-language-into-action/ systems-at-ricm/ibm-mtst-system

85. Franzen, C. OpenAI powers a robot. 107. Mehra, V., and Hasegawa, R. (2023, October 3) Supporting power grids
with demand response at Google data centers. Google Cloud:
https://cloud.google.com/blog/products/infrastructure/using-demand-re-
96. Vuong, Q., and Sanketi, P. (2023, October 3). Scaling up learning across sponse-to-reduce-data-center-power-consumption 117. Computer History Museum. (2012, January 3). Vintage Steve Jobs footage
many different robot types. Google DeepMind: https://deepmind.google/ on Apple. YouTube: https://www.youtube.com/watch?v= GfxxRKBgos8
86. $4.2 Billion Invested: Robotics Startups See Funding Surge. (2024). discover/blog/scaling-up-learning-across-many-different-robot-types/
Founders Network: https://foundersnetwork.com/blog/robotics-start-
ups-see-funding-surge/
108. Lewis, M. (2024, March 5). Amazon just bought a 100% nuclear-powered
data center. Electrek: https://electrek.co/2024/03/05/amazon-just- 118. Blitz, M. (2019, April 4). Later, Navigator: How Netscape Won and Then Lost
97. Chen, P. (2024, January 8). The Future of Robotics: Robotics Foundation bought-a-100-nuclear-powered-data-center/ the World Wide Web. Popular Mechanics: https://www.popularmechanics.
Models and the role of data. Covariant.ai: https://covariant.ai/insights/the- com/culture/web/a27033147/netscape-navigator-history/
87. Gao, J., Sarkar, B., et al. (2023, September 5). Physically Grounded future-of-robotics-robotics-foundation-models-and-the-role-of-data/
Vision-Language Models for Robotic Manipulation. Stanford.edu:
https://iliad.stanford.edu/pg-vlm/
109. Kimball, S. (2024, September 20). Constellation Energy to restart Three
Mile Island nuclear plant, sell the power to Microsoft for AI. CNBC: 119. Dimitri, C., Effland, A., and Conklin, N. (2005). The 20th Century Trans-
98. NVIDIA Isaac Sim. (n.d.). NVIDIA Developer: https://developer.nvidia.com/ https://www.cnbc.com/2024/09/20/constellation-energy-to-restart- formation of U.S. Agriculture and Farm Policy. United States Depart-
isaac/sim three-mile-island-and-sell-the-power-to-microsoft.html ment of Agriculture: https://www.ers.usda.gov/webdocs/publica-
88. Chiang, H., Xu, Z., et al. (2024, July 12). Mobility VLA: Multimodal Instruc- tions/44197/13566_eib3_1_.pdf
tion Navigation with Long-Context VLMs and Topological Graphs. arXiv.
org: https://arxiv.org/pdf/2407.07775

99. Open X-Embodiment: Robotics Learning Datasets and RT-X Models (n.d.). 110. Clifford, C. (2023, May 10). Microsoft agrees to buy electricity generated
GitHub: https://robotics-transformer-x.github.io/ from Sam Altman-backed fusion company Helion in 2028. CNBC: https:// 120. Heaven, W. (2020, July 20). OpenAI’s new language generator GPT-3 is
www.cnbc.com/2023/05/10/microsoft-agrees-to-buy-power-from-sam- shockingly good—and completely mindless. MIT Technology Review:
89. Driess, D., and Florence, P. (2023, March 10). PaLM-E: An embodied multi- altman-backed-helion-in-2028.html https://www.technologyreview.com/2020/07/20/1005454/openai-ma-
modal language model. Google Research: https://research.google/blog/ chine-%20learning-language-generator-gpt-3-nlp/
palm-e-an-embodied-multimodal-language-model/
100. Harris, M. (2022, August 3). The Radical Scope of Tesla’s Data Hoard. IEEE
Spectrum: https://spectrum.ieee.org/tesla-autopilot-data-scope
111. Industrial Revolution. (n.d.). Britannica: https://www.britannica.com/
event/Industrial-Revolution 121. Spataro, J. (2023, March 16). Introducing Microsoft 365 Copilot – your co-
90. Chu, J. (2024, March 25). Engineering household robots to have a little pilot for work. Microsoft: https://blogs.microsoft.com/blog/2023/03/16/
common sense. MIT News: https://news.mit.edu/2024/engineering-house- introducing-microsoft-365-copilot-your-copilot-for-work/
hold-robots-have-little-common-sense-0325 101. Tesla AI. (n.d.): https://www.tesla.com/AI
Technology Vision 2025 | AI: A Declaration of Autonomy References 66

References

122. Accenture Report Finds Perception Gap Between Workers and C-suite 131. Heater, B. (2024, April 25). Sanctuary’s new humanoid robot learns faster
Around Work and Generative AI. (2024, January 16). Accenture: https:// and costs less. TechCrunch: https://techcrunch.com/2024/04/25/sanctu-
newsroom.accenture.com/news/2024/accenture-report-finds-perception- arys-new-humanoid-robot-learns-faster-and-costs-less/
gap-between-workers-and-c-suite-around-work-and-generative-ai

132. Sanctuary AI Deploys First humanoid General-Purpose Robot Commercial-

123. AI Can Make Workers Happier. (2024, July 2). The Global Recruiter: ly. (2024, March 7). Sanctuary AI: https://www.sanctuary.ai/blog/sanctu-
https://www.theglobalrecruiter.com/ai-can-make-workers-happier/ ary-ai-deploys-first-humanoid-general-purpose-robot-commercially

124. Toyota Research Institute Unveils New Generative AI Technique for Vehicle 133. Resendes, S. (2022, January 13). The Hospitality Quotient: Danny Meyer’s
Design. (2023, June 20). Toyota: https://pressroom.toyota.com/toyota-re- Hiring Formula For Building a Brand. Lightspeed: https://www.lightspeed-
search-institute-unveils-new-generative-ai-technique-for-vehicle-design/ hq.com/blog/the-hospitality-quotient/

125. Google Cloud. (2024, April 9). How Wayfair is using Gemini CodeAssist 134. Moody, K. (2024, July 29). College education may not be preparing
to increase developer productivity. YouTube: https://www.youtube.com/ employees for generative AI. Higher Ed Dive: https://www.highereddive.
watch?v=VwZNWqh5ZWU com/news/college-education-not-preparing-employees-for-genera-
tive-ai/722327/

126. First Generative AI Drug Begins Phase II Trials with Patients. (2023, July 1).
Insilico.com: https://insilico.com/blog/first_phase2

127. Dunn, J. (2024, May 16). Camunda simplifies process automation with new
AI-powered natural language features. CIO.com: https://www.cio.com/ar-
ticle/2110581/camunda-simplifies-process-automation-with-new-ai-pow-
ered-natural-language-features.html

128. Lagerkrans-Pandey, S., Bolaños, M., et al. (2024, June 24). Introduction to
Semantic Kernel. Microsoft: https://learn.microsoft.com/en-us/seman-
tic-kernel/overview/

129. Kerner, S. (2023, August 8). Kognitos goes self-service with business auto-
mation powered by generative AI. VentureBeat: https://venturebeat.com/
automation/kognitos-goes-self-service-with-business-automation-pow-
ered-by-generative-ai/

130. Burger, D. (2023, September 25). AutoGen: Enabling next-generation

large language model applications. Microsoft: https://www.microsoft.
com/en-us/research/blog/autogen-enabling-next-generation-large-lan-
guage-model-applications
Technology Vision 2025 | AI: A Declaration of Autonomy 67

About Accenture

Accenture is a leading global professional services company that helps the world’s leading businesses, governments and
other organizations build their digital core, optimize their operations, accelerate revenue growth and enhance citizen
services—creating tangible value at speed and scale. We are a talent- and innovation-led company with approximately
799,000 people serving clients in more than 120 countries. Technology is at the core of change today, and we are one
of the world’s leaders in helping drive that change, with strong ecosystem relationships. We combine our strength in
technology and leadership in cloud, data and AI with unmatched industry experience, functional expertise and global
delivery capability. Our broad range of services, solutions and assets across Strategy & Consulting, Technology,
Operations, Industry X and Song, together with our culture of shared success and commitment to creating 360° value,
enable us to help our clients reinvent and build trusted, lasting relationships. We measure our success by the 360° value
we create for our clients, each other, our shareholders, partners and communities. Visit us at accenture.com.

Disclaimer: The material in this document reflects information Accenture does not provide legal, regulatory, audit,
Accenture Research available at the point in time at which this document was prepared or tax advice. Readers are responsible for obtaining
as indicated by the date provided on the front page, however the such advice from their own legal counsel
global situation is rapidly evolving and the position may change. or other licensed professionals.
Accenture Research creates thought leadership about the most pressing business issues organizations face. This content is provided for general information purposes only,
Combining innovative research techniques, such as data-science-led analysis, with a deep understanding of does not take into account the reader’s specific circumstances, This document refers to marks owned by third parties.
All such third-party marks are the property of their respective
industry and technology, our team of 300+ researchers in 20 countries publish hundreds of reports, articles and is not intended to be used in place of consultation with our
professional advisors. Accenture disclaims, to the fullest extent owners. No sponsorship, endorsement or approval of this content
and points of view every year. Our thought-provoking research developed with world leading organizations permitted by applicable law, any and all liability for the accuracy and by the owners of such marks is intended, expressed or implied.
helps our clients embrace change, create value and deliver on the power of technology and human ingenuity. completeness of the information in this document and for any acts or
Copyright © 2025 Accenture. All rights reserved.
Visit us at accenture.com/research. omissions made based on such information.
Accenture and its logo are registered trademarks of Accenture.