Contents

4
1 Introduction...............................................................................................................

2 Getting Started.......................................................................................................6
Strong Communication Skills
Creativity and Critical Thinking
Logical Reasoning Skills
Start With Simple, Unstructured Conversations

3 Making GenAI Conversations More Effective........................................... 10

Key Strategies for Conversation
Universal Rules of Conversation
Prompting Patterns
Prompting Techniques

4 Automating, Assisting and Augmenting Project Work............................. 17

Automation Prompts Save Time
Assistance Prompts Fill Knowledge Gaps
Augmentation Prompts Help You Wrestle With Complex Challenges

5 Further Guidance................................................................................................ 24
Excellent Prompting References
Continue Your PMIxAI Journey

6 Appendix: Plug-and-Play Prompt Ideas for Project Professionals..... 26

Example Automation Prompt: Meeting Minutes Summary
Example Assistance Prompt: Generating a Risk Matrix
Example Augmentation Prompt: Framing a Decision

7 Acknowledgments................................................................................................ 31
A
s a project manager, I’ve always envisioned a day when I could “talk to
a machine” — not in the command and control sense that we all do
when using our smartphones, but in a more conversational manner
of talking about issues, problems and different perspectives of my daily
work. I imagined discussions on project management topics such as crafting
a project charter, deliberating on critical project goals or brainstorming
potential risks. I am thrilled that this day has arrived!

At the intersection of human language and computational power lies a

transformative domain known as prompt engineering. This discipline transcends
mere technical competencies; it is a language that merges human intellect,
ingenuity and creativity with the full potential of artificial intelligence (AI).

At Project Management Institute, our PMIxAI initiative includes several efforts

to understand and harness the impact and potential of AI and generative
AI (GenAI) tools within project management. For example, PMI Infinity™, our
AI-powered learning copilot, can help you take your project management
practice to the next level.

To complement these advancements, we’ve compiled this comprehensive

report on prompt engineering. Our main objective is to equip project
professionals with the knowledge to navigate and leverage GenAI tools,
thereby enhancing their work and contributing to optimized project success.

This report delves into prompt engineering techniques and provides

recommendations drawn from technical sources and the practical insights of
global experts. It features contributions from our community of professionals
and industry leaders who are exploring the vast potential of GenAI in project
management. The future of project management is full of possibilities:
Mastering the art of communicating with machines will ensure we are equipped
to seize the opportunities of this new technological era. Through prompt
engineering, project professionals can elevate their power skills, expand their
business acumen and revolutionize their ways of working.

I invite you to continue reading to uncover the possibilities of prompt

engineering techniques and to better understand the impact this competence
can provide to your everyday life as a project professional. Share this
knowledge with your teams and organization, empowering them to harness the
transformative benefits of GenAI.

Edivandro Carlos Conforto, Ph.D.

Director of Thought Leadership, PMI

Building and Leading High-Performing Teams 3

1.0

Introduction
The adoption of GenAI tools based on large language models
(LLMs) has driven the rise of a new digital competence — prompt
engineering — that may be unfamiliar to project professionals. Since
the release of ChatGPT in November 2022, prompt engineering for
GenAI has become recognized as a necessary digital skill. Baidu CEO
Robin Li 1 predicts, “in the next decade, half of all jobs worldwide will
revolve around prompt engineering.”

What are prompts and what is prompt engineering? According

to MIT, 2 “Prompts are your input into the AI system to obtain
specific results. In other words, prompts are conversation starters:
what and how you tell something to the AI for it to respond in
a way that generates useful responses for you.” By extension,
prompt engineering is the process of designing natural language

4
1.0 INTRODUCTION

specifications of a task, which are used to will need to understand what prompt
condition the LLM at inference time. “One engineering is and how to develop and apply
reason this latest generation of AI tools has this skill to different project management
so much promise is that they make it so tasks.
people can now interact with our computers
in the same way we’ve interacted with other Why? “It’s very possible to be having a
humans for millennia: natural language,” as conversation with your AI tool, while not
Jaime Teevan 3 of Microsoft explains it. actually getting the answer you want
because you didn’t ask it the correct
The Generative AI Overview for Project question,” explains Bart Gerardi, senior
Managers course and the latest thought manager, software engineering, Redfin. “The
leadership research from PMI have AI tool may know the answer, but you don’t
highlighted how effective prompts can always know what to ask for and how to ask
generate more accurate and relevant for it.”
outputs from LLMs and popular generative
AI tools, e.g., ChatGPT (OpenAI), Gemini
(Google), Claude 2 (Anthropic) and Copilot
(Microsoft). Project professionals seeking
to leverage the capabilities of GenAI tools

SIDEBAR

Prompting: A Key Digital Competence

“Seventy-four percent of executives believe that generative AI will benefit their employees,”
according to a LinkedIn report. 4 Further, the report indicates a “substantial increase” in searches
for terms such as “ChatGPT,” “prompt engineering,” “prompt crafting,” “Microsoft Copilot,” and
“generative artificial intelligence” on member profiles globally.

According to a BCG study, 5 nearly all participants (around 90%), irrespective of their baseline
proficiency, produced higher-quality results when using GPT-4 for the task. Further, the variance
in performance dramatically reduced, meaning a much higher share of participants performed
at or very close to the average level, substantially increasing the average performance levels of
the entire group.

1 Fu, Z. (2023, May 27). Baidu’s Robin Li: all products will be remade to become AI-Native. Pingwest.
2 MIT Management. (n.d.). Effective Prompts for AI: The Essentials. MIT Sloan Teaching & Learning Technologies.
3 Teevan, J. (2023, December 15). To Work Well with GenAI, You Need to Learn How to Talk to It. Harvard Business Review.
4 LinkedIn Economic Graph. (2023, November). Future of Work Report: AI at Work. LinkedIn.
5 Candelon, F., Krayer, L., Rajendran, S., & Martínez, D. Z. (2023, September 21). How People Can Create—and Destroy—Value with Generative AI.
BCG Global.

5
2.0

Getting Started
Prompting is your opportunity to move from a “shaper of project
work” to a value creator. Treat it that way and dive right into the
experience.

First, trust yourself and resist viewing prompting as complicated

or daunting. Many of the same skills that make you a great project
manager will serve you well here.

6
2.0 GETTING STARTED

2.1 Strong Communication Skills

Clear and precise communication is essential for managing projects and formulating prompts. In both
cases, you articulate project objectives and requirements effectively, thus aligning both people and
GenAI tools with the specific needs and goals of the project. Strong communication skills can help you
— both with humans and AI — to:

• Reduce the likelihood of misinterpretation.

• Provide context, specific constraints and desired outcomes of the request.

• Explain technical concepts to enhance problem-solving capabilities.

• Navigate specialized terminology and jargon, aligning it with the specific terminology used in the
project.

Partnering with AI can further strengthen project managers’ skills because they can assess, refine and
polish the communication options and approaches the GenAI provides.

2.2 Creativity and Critical Thinking

Creativity, coupled with critical thinking, aids in identifying unconventional solutions. Partnering with AI is
similar to partnering with humans when addressing complex problems that may require knowing what
questions to ask, thinking outside the box and exploring alternative approaches. Both humans and AI
can come together to:

• Brainstorm and then narrow down the focus to ensure we ask the right questions. GenAI can
offer diverse perspectives on a topic, help consider different angles and uncover overlooked
questions in a rapid-fire way, but it’s only the human’s critical thinking that can guide this process
— and shape and refine it.

• Experiment and refine the approach iteratively, continually improving prompt effectiveness. GenAI
can help test your ideas and produce additional options for discovery.

AI tools can help build project management skill sets by making unforeseen connections, filling gaps in
concepts and generating new ideas to consider, which can be refined and combined based on human
knowledge and experience.

2.3 Logical Reasoning Skills

Logical reasoning leads to clear instructions, a logical progression of ideas, identification of key
relationships and the minimization of ambiguity. Reasoning skills also help with effective problem
decomposition and structured query formulation. Logic makes it possible to:

• Present ideas in a coherent manner by following a structured flow of information.

• Break down complex problems into smaller, meaningfully connected components. This
decomposition allows both humans and AI tools to address each component independently, aiding
in a more focused and accurate response.

7
2.0 GETTING STARTED

• Identify and express key relationships within complex problems. Clearly articulating these
relationships helps humans and sophisticated AI tools to grasp the interdependencies and
contextual information necessary for generating meaningful responses.

• Ask the right questions. This learned skill is important when dealing with complex inquiries. A
well-structured question helps humans and their GenAI tools to comprehend and respond more
effectively.

Project managers can strengthen their problem-solving abilities by reinforcing the value of logical
approaches based on improved outputs. Applying prompting patterns and techniques also helps hone
the project managers’ approaches to regular human interactions, offering logically structured ways to
express concepts.

2.4 Start With Simple,

Unstructured Conversations
Talking to a GenAI assistant takes some getting
used to. You begin by simply asking questions.
However, it’s best to avoid structuring your “A good way to get over
queries the way you would for a search engine; the resistance could be to
don’t just ask a question and leave it there. You approach it conversationally.
should prepare to build out at least three to four Tell your GenAI tool: ‘I’m
exchanges in every conversation to help arrive at going to call you Charles,’
a style of interaction that leads to the answers and then talk to Charles
you want. Conversation starters can begin the about your project like you
process, but remember to ask deeper questions would to a team member.”
based on the GenAI responses. Starters include:
- Nathan Rischling, program manager,
• I am a project manager and am in the Sigma Defense Systems, USA
planning phase of my project. I don’t know
which project management approach to use.
What are the key considerations for selecting
one?

• I am a project manager with some

experience. I am frustrated because things
were proceeding well in the first iterations,
but now things have changed, and our
deliverables are lacking the agreed-upon
quality. They are also arriving late. What can
I do?

• I am an experienced project manager and

am good at engaging stakeholders. But in
my current project, the deliverables don’t
satisfy them. We have even shifted from a
predictive to an agile approach, but they
keep on changing the project scope. What
should I do?

• I am a new project manager. I need to

evaluate which tools I can use to measure
stakeholder satisfaction with the project.
Give me some names with a detailed
breakdown of why I should use them.

8
2.0 GETTING STARTED

SIDEBAR

Resilience and Curiosity Are Important for Learning Prompt

Engineering
While resilience and curiosity may not directly influence the content of prompts, they are key
attributes that help project managers navigate the complexities of prompt engineering and AI
interactions. Here’s why:

• Prompt engineering is often an iterative process. Resilience enables project professionals to

persist in refining prompts through multiple iterations.

• There is no single way to engineer prompts. Curiosity drives project managers to

proactively explore different prompt engineering strategies, which is crucial for discovering
innovative approaches, experimenting with new techniques and optimizing prompt
formulations to improve the effectiveness of interactions with the AI tool.

“Think of prompt engineering

like crafting clear messages
in project management. Be
resilient — keep refining
and improving your messages
persistently. Stay curious
— explore different ways to
make your messages better
and more innovative. For
example, instead of saying
‘Discuss issues,’ say ‘Share
project challenges’ for a
more relevant output. Keep
trying new ways, like changing
‘Provide updates’ to ‘Detail
achievements and setbacks.’
Test your messages in context,
like moving from ‘Next steps?’
to ‘Identify critical tasks
for the upcoming sprint.’
Stay curious, and your
messages, whether for projects
or chatbots, will be more
effective and versatile.”
- Subhankar Modak, PMO Program Manager,
Accenture, India

9
3.0

Making GenAI
Conversations
More Effective
GenAI performs best when talked to like a human. But it’s a human
with a specific personality that you need to adjust for. There are five
key strategies and three universal rules to guide you through that
adjustment. The AI does better if the conversation is well structured.
Prompting patterns describe how your conversation should flow to
get the responses you want. Finally, there are prompting techniques
that get you to the best and most useful response possible.

10
3.0 MAKING GENAI CONVERSATIONS MORE EFFECTIVE

3.1 Key Strategies for Conversation

The entire point of prompt engineering is to create prompts for unique use cases and iterate them
as required. Project professionals can utilize this quick checklist of strategies to harness their prompt
engineering:

1. Use a “diverge and converge” approach. Start with a broad, open-ended question to utilize
the GenAI tool for its exploration capabilities. Once the tool provides a general framework,
converge on each element. Use the following steps to generate more effective answers. Project
teams can also design a sequence of prompts to address distinct outputs and objectives.

2. Provide more context than you would to a human. Detail the problem or question well, and
be specific about things such as industry, region, type of project and the task. Outline the project’s
background, describe the person using the GenAI tool (like a project manager) and say how and
where the answer will be used. The granularity of the input will be directly proportional to the utility
of the output received.

3. Give examples. Embed specific examples in the input text to aid the LLM in understanding
the desired output format or preferred answer. It can be something as simple as: “Use this as an
example ...’’ This proves especially beneficial for tasks demanding a particular structure or format,
like generating code or crafting specific types of text. As OpenAI 6 says in its GPT-4 guide: “The less
the model has to guess at what you want, the more likely you’ll get it.”

4. Make it a conversation, not a command. Per the OpenAI 7 GPT-4 guide: “These models can’t
read your mind. If outputs are too long, ask for brief replies. If outputs are too simple, ask for
expert-level writing. If you dislike the format, demonstrate the format you’d like to see.” Interact
with the LLM like it’s an intern who needs constant instructions. Note what types of prompts give
the best results and refine the input accordingly.

5. Incorporate a “reliability check.” Include a request for references and sources in the ask.
Even though some LLMs don’t browse the web in real time, this can still prompt them to provide
the name of a source that can be verified via a simple online search. It also prompts the model to
assess the output as a “sanity check” and helps ensure that the generated content aligns with the
desired accuracy and reliability standards.

“Nobody wants a quarter-inch drill. What they want is a

quarter-inch hole. We’re so accustomed to asking for the
drill, but we’re not accustomed to asking for the actual
outcome that we want. To use GenAI properly, we need a
shift in how project professionals think. They should be
asking for their outcomes.”
- Bart Gerardi, senior manager, software engineering, Redfin, USA

6 OpenAI. (n.d.). Six Strategies for Getting Better Results. OpenAI.

7 OpenAI. (n.d.). Six Strategies for Getting Better Results. OpenAI.

11
3.0 MAKING GENAI CONVERSATIONS MORE EFFECTIVE

3.2 Universal Rules of Conversation

As mentioned above, GenAI has a specific personality. It helps if you think of the AI tool as a smart
generalist you know — one who has questionable ethics and can’t keep a secret.

Three simple rules apply to conversations with GenAI:

1. Don’t rely on the LLM’s domain knowledge too much. While LLMs like GPT-3.5/4.0 possess
broad general knowledge, they may lack specifics and contextual information about certain
industries or projects. Use the LLM’s domain knowledge as a starting point and use your human
expertise to counter-check what it tells you.

2. Don’t share confidential or sensitive data with the tool. Unless the LLM you are working on
exists in a controlled IT environment of your organization, assume that it will share everything you
tell it with everyone else using it. Avoid including sensitive or confidential information in prompts
to uphold privacy and data security. Craft prompts that focus on extracting insights or actions
without compromising proprietary or confidential details, ensuring compliance with privacy
regulations and safeguarding project integrity.

3. Be mindful of ethical guidelines. Many organizations have already started putting ethical
guidelines and considerations in place. Without those guidelines, you should assume that AI
tools have the potential to generate inappropriate or harmful content. Be mindful of ethical
considerations and ensure that prompts align with ethical standards and professional conduct.

“Think of generative AI as
the starting point. Remember:
You have to use your human
intelligence on top of the
generative AI’s intelligence.”
- Pamela Krzypkowska, director of research
and innovation, Ministry of Digital Affairs,
Poland

12
 3.0 MAKING GENAI CONVERSATIONS MORE EFFECTIVE

3.3 Prompting Patterns

Table 1 provides several common types of prompt patterns, along with recommendations on when to
use them and how to structure the prompts. The information and examples presented here are not
exhaustive and should be adapted to the specific project context to provide better outputs.

Table 1. Patterns of Prompts to Support Prompt Engineering (adapted from White et al., 20238

Prompt pattern Use this to … Basic structure Prompt examples

Persona Make your generative “From now on, act as X …” Act as a stakeholder of
pretraining transformer the project. Perform an
(GPT) an expert. Use “Act as persona X …” assessment of the project
this pattern for a charter (or deliverable)
conversational interaction “Create outputs that and provide a list of
that can frame the output persona X would create …” recommendations to
from the perspective of improve scope clarity and
a specific role within the completeness.
project environment, or
ask the tool to respond
as a nonhuman entity,
e.g., project management
software.

Question refinement Drill down. Use this “Within scope X, suggest Within the scope of
pattern to create and a better version of the creating a cost-benefit
refine questions about question to use instead …” analysis of outsourcing
specific topics or domains (total cost: US$60,000)
of a project or industry, or versus in-house hiring
when there is not clarity (total cost: US$70,000),
about the problem to be suggest a better version
solved. of the question to be used.

Flipped interaction Practice for a human “I would like you to ask me I would like you to ask
conversation. Use this questions to achieve X …” questions to help with
pattern to simulate a my preparation for a
conversation with project “You should ask questions project update with my
stakeholders, team until condition X is met or stakeholders [provide the
members, clients, users or to achieve goal Y …” context of the project].
another project manager Ask me questions one at a
in order to identify gaps “Ask me questions one at time so you can build the
and audit deliverables. a time …” next question based on my
responses.

Alternative approaches Weigh options. “Within scope X, if I am a construction

Use this pattern when there are alternative project manager
there is a need to identify ways to accomplish the experiencing a delay in
different approaches, same thing, list the best completion of some tasks.
evaluate pros and cons, alternate approaches, I want to identify the root
compare approaches and compare/contrast the cause(s) of the problem.
define which approach pros and cons of each Suggest at least two
would be better suited approach …” options to deal with the
to solve a problem in the problem(s) identified and
project. compare the pros and
cons of each. Present the
results in a table.

8 White, J. (2023, February 21). A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT arXiv.org.

13
Prompt pattern Use this to … Basic structure Prompt examples

Cognitive verifier Simplify a complicated “When you are asked a A software development
question. Use this pattern question, follow these project has a US$100,000
when you need to combine rules …” budget and a timeline of
two or more questions six months. After three
related to different “Generate a number months, the actual cost
dimensions of the project of additional questions (AC) is US$60,000 and the
and the task becomes too that would help to more project is 50% complete.
complex and confusing. accurately answer the What are the risks of not
To ensure the large question …” completing the project on
language model (LLM) will time and on budget? Am
generate the best output, “Combine the answers to I over or under budget at
use this pattern to guide the individual questions to this point? Am I behind or
the tool to help break produce the final answer ahead of schedule? What
down the question into to the overall question …” is the earned value (EV),
more specific requests estimate to complete
and perform the tasks (ETC) and value-added
accordingly. control? Combine the
answers to the individual
questions to produce the
final answer to the first
question.

Fact check list Build a fact base. Use this “Generate a set of facts I need to identify the most
pattern when creating that are contained in the common challenges based
outputs for the project output …” on customer feedback
that are based on external and industry information
information, like industry “The set of facts should to create a list of user
perspectives, domain- be inserted in a specific stories to develop a
specific information, data point in the output …” mobile app that allows
or definition of concepts frequent fliers to manage
to avoid factually their flights. Generate
incorrect outputs or a list of the 10 most
information. common frequent flier
challenges in the airline
industry based on facts.
The set of facts should
be inserted in a specific
point in the output and
correctly show links to
sources.

Visualization generator Build a [diagram name]. “Generate an X that I I want to generate a

Use this pattern when can provide to tool Y to prompt that I can provide
writing input text to a visualize it …” to tool [X] to create a
text-to-image tool; for diagram to visualize the
example, using DALL-E (or 10 most common sources
ChatGPT Plus) to generate of project scope creep to
charts, diagrams and use in my project kickoff
images to support project presentation with a new
work and deliverables. client.

Reflection Explain an answer. Use “Explain the reasoning and I need to provide feedback
this pattern when there is assumptions behind your to a project team member
the need to have a clear answer …” who made a mistake with
rationale about an output, financial implications
explanation or reasoning. “Whenever you generate to the organization.
It can be used for risk an answer …” They are very talented
analysis, project decision- and my intent is to help
making recommendations with their development
and feedback and and growth. Mistakes
assessment sessions. happen, and I don’t want
this to become a blocker
for their creativity and
experimentation mindset.
Explain the reasoning and
assumptions behind your
recommendation on how
to approach this meeting
with them.

14
 3.0 MAKING GENAI CONVERSATIONS MORE EFFECTIVE

3.4 Prompting Techniques

Understanding prompting techniques can empower project professionals to craft prompts that range
from providing minimal information to offering contextual guidance, allowing for nuanced control over
the creative and analytical capabilities of the model.

Table 2 details three popular prompting techniques and provides examples of how project
professionals can use them.

Table 2. Most Common Prompting Techniques (Source: IBM)9

Prompting Level of
Description Example
technique complexity

Zero-shot prompts Think of this as the closest Low “What are the key
match to a simple query elements of a project
on a search engine. Zero- charter?”
shot prompting tests the
large language model’s
(LLM) ability to produce an
accurate output, without
relying on prior examples.
This is what an average
user who is not explicitly
training the LLM for a task
uses.

Few-shot prompts Think of this as telling Medium “For a remote team

a search engine, “I’m spread across time zones,
interested in these what are some effective
specific topics. Now refine project communication
your search based on methods?”
these examples.” Few-
shot prompting needs
the user to train the LLM
with a few examples and
sample outputs, such that
the LLM has context to
draw on and can better
understand the expected
outcome.

Chain-of-thought (CoT) Think of this as an AI High “How does the

prompts version of a dynamic identification of project
brainstorming session — risks influence the
one that uses previous development of risk
thoughts and information. response strategies,
With chain-of-thought and how can these
prompting, users need strategies be integrated
to break down a complex into the overall project
task into intermediate management plan?”
steps or “chains of
reasoning.” They then need
to provide step-by-step
reasoning for the model to
follow and help it create
more accurate outputs.

9 IBM. (n.d.). What is prompt engineering? IBM.

15
3.0 MAKING GENAI CONVERSATIONS MORE EFFECTIVE

SIDEBAR

3.5 Familiarity With LLMs Can Give You an Edge

Knowing the strengths and weaknesses of LLMs empowers project managers to use these
models strategically for problem-solving and creative tasks. By tailoring prompts to tap into the
model’s capacity for generating diverse and contextually relevant content, project professionals
can derive valuable insights, suggestions or creative solutions for project challenges.

• Different LLMs and GenAI tools have specific architectures and capabilities. Understanding
these better allows project managers to align their prompt engineering strategies with
the underlying architecture of the AI tool, optimizing the utilization of its strengths and
addressing potential limitations.

• Familiarity with LLMs aids in interpreting and evaluating the outputs generated by these
models. Project professionals can discern the model’s thought process, identify potential
biases and assess the reliability of responses. This understanding is crucial for making
informed decisions based on the insights provided by GenAI.

These models evolve rapidly over time, so it’s important to keep abreast of recent changes.

“Being aware of the basics

of LLMs when using GenAI
models can give users a very
important edge — knowing
elements like which of the
available models is connected
to the internet and which
operates like a stand-
alone box, and knowing the
strengths and weaknesses of
each model, can be crucial to
understand which GenAI tool
is appropriate for which use
case, and what the limitations
would be.”
- Pamela Krzypkowska, director of research
and innovation, Ministry of Digital Affairs,
Poland

16
4.0

Automating,
Assisting and
Augmenting
Project Work
The PMI Thought Leadership report, Shaping the Future of Project
Management, describes a three-tier model for applying GenAI.
Effective prompting is what makes each of these solution tiers
come to life.

17
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

High

Project business
case creation
How complex is the task?

Large dataset
analysis Risk analysis
Project decision
making

Lessons learned
summary Cost and schedule
estimation

Cost-benefit
Report analysis
generation

Meeting notes
summary
Automating Assisting Augmenting
Low

Machine Who does the work? Human

Figure 1: Sample Project Management Tasks That Can Be Automated, Assisted and Augmented
With GenAI

The scope of each of these tiers is described in Figure 1, but a simple way to think about them is:

• Automating can be thought of as “word crunching.” GenAI uses an advanced version of

autocomplete to create, structure and summarize common project management documents. An
example of automation is generating meeting minutes, which is typically done almost entirely by
the machine and then lightly checked by the project manager. Automation helps you increase
speed and productivity.

• Assisting provides highly targeted knowledge management. For example, the AI tool will give
you all of the common risks or potential costs and benefits related to a given project situation.
Assistance helps you fill gaps and include things you may have missed.

• Augmenting is focused on reasoning. When you have a layered challenge, like building a business
case or brainstorming ideas, GenAI can serve as your thought partner, working through the
component parts of a problem until you have solved it. Augmentation helps you step through
tough problems without getting stuck.

18
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

4.1 Automation Prompts Save Time

Automating project tasks using prompts in GenAI helps you create, structure and summarize common
project management documents.

With automation, the following prompt patterns are very useful:

• Persona. Ask the AI tool to play the role of an expert in project management for the industry
and project type you are working on. This allows the tool to deliver a more nuanced version of the
report you are trying to generate.

The following prompting techniques are also helpful.

• Few-shot prompts. If you have an example of the document you are trying to create that
illustrates the desired format and includes the right data elements, provide it to the AI tool as part
of the prompt. This will get you much closer to the result you seek.

Beyond meeting minutes, based on predefined parameters, GenAI can also automate:

• Regular status communications to various stakeholder groups.

• Reporting of activities, performance metrics and more.

ACTION:
An example of an automation
prompt is offered in the Appendix:
“The goal of GenAI is to Summarizing Meeting Minutes. You can
take a project professional’s easily apply this in an upcoming meeting,
mundane tasks and make where you have either recorded a transcript
them easy, like ‘here you or taken robust notes. This should save a
go, this is done.’ Learning great deal of time when reorganizing the
material into a document that is useful to
to leverage it ultimately stakeholders.
increases the value of the
project professional to the
TIP: Prompting for automation typically
organization, because not only
involves documents you create on a regular
is it saving the professional’s basis. Early in the process, your output will
precious time, but it is also not be perfect. The temptation will be to
empowering the professional to simply reuse the prompt as is and clean up
focus on strategic aspects of each document. However, if you are able
the job that no one else can to refine the prompt and solve your issues
do.” that way, you’ve saved “future you” even
more time and effort. This will also make you
- John Signo, Vice President, VP DMT Delivery better at prompting in general.
Lead, Bank of America, USA

19
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

4.2 Assistance Prompts Fill Knowledge Gaps

Research10 shows that 94% of the time, engagement with AI (including GenAI) leads to asking a
wider range and variety of questions than the respondents would have otherwise. This, in turn,
leads to exploring ideas and possible solutions they may not have considered, likely leading to better
performance.

Gaining GenAI assistance on project tasks using prompts helps avoid potential gaps in sets of options
that may occur when generating a document or simply brainstorming. Further, it can be your secret
weapon when venturing into new territory. AI tools can provide quick summaries, present unfamiliar
concepts and explain existing perspectives in any domain you might be new to.

Common examples of this include:

• Gathering and ranking risks for a risk matrix/assessment.

• Gathering and assessing costs and benefits for a cost-benefit analysis.

With assistance, the following prompt patterns are very useful, in addition to the persona prompt
pattern discussed earlier:

• Flipped interaction. Ask the tool: “What additional questions would you ask in this situation?” This
prompt can yield avenues of inquiry you had not previously considered, particularly toward the end
of the conversation.

• Alternative approaches. Different approaches can help you identify new ways of looking at a
problem, which can provide a “devil’s advocate” position.

• Question refinement. Narrowing down the focus can lead to the right set of answers, simply by
zeroing in on the root of the problem.

In addition to the few-shot prompting technique, the following prompting techniques are also helpful:

• Chain-of-thought (CoT). As the AI tool surfaces new options, clarification may be required. You
may want to understand the options in order of priority, or you may want to understand the
entire option set as an interrelated system. Finding the right information requires drilling down by
asking more questions that follow a logical line of inquiry.

“I imagine a future state where project professionals have

GenAI looking at project schedules, looking at plans, maybe even
looking at a project board. And the AI can identify risks or it
can identify scheduling challenges or it can find mistakes. You
have a dependency you don’t know about; you have a risk you don’t
know about; you have something you’re not highlighting — GenAI
can help you spot that very quickly. I don’t think GenAI can run
a project, but I think as a reviewer of project work, it can find
errors in a much faster way than humans can.”
- Bart Gerardi, senior manager, software engineering, Redfin, USA

10 Gregersen, H., & Bianzino, M. N. (2023, May 26). AI Can Help You Ask Better Questions — and Solve Bigger Problems. Harvard Business Review.

20
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

Beyond risk management and cost-benefit

analysis, based on predefined parameters, GenAI
can also assist with suggesting options for: ACTION:
• Project planning and scheduling An example of an assistance
prompt is offered in the Appendix:
• Resource management Generating a Risk Matrix. You can easily
apply this to your current project, creating
• Task prioritization and assignment a new document or cross-checking one you
have already prepared. This could yield fresh
• Communication and collaboration perspectives you may not have considered.

• Quality assurance TIP 1: Effective prompting involves more

than simply cutting and pasting a simple
• Change management prompt. Refining the focus and asking more
detailed questions can give you better
• Knowledge management and learning information, increasing the quality of your
final work product.
• Stakeholder analysis and engagement
TIP 2: A subject matter expert should always
• Financial management validate assisted work products. Blindly
accepting an AI output is not advisable.
• Risk monitoring and management

• Brainstorming exercises

AI tools can do this by generating lists or

developing complete documents.

21
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

4.3 Augmentation Prompts

Help You Wrestle With
Complex Challenges
GenAI augmentation using prompts makes
people better at dealing with complexity. The AI “Sometimes my brain is not
“augments” your thinking and capabilities when enough. When I am faced with
working through layers of interrelated details, complex decision-making, that
helping you arrive at higher-quality solutions. requires a sophisticated
Common examples of this include:
and creative approach, I
• Building business cases.
sometimes use GenAI to
augment my decision-making.
• Framing complex decisions. I use GenAI help me see
things from multiple angles,
In the augmentation tier, persona, flipped come up with innovative
interaction, alterative approach and question approaches and even make
refinement prompt patterns (discussed earlier) intelligent suggestions.”
are very useful, in addition to the following:
- Yuliya Zhevno, project and change
• Cognitive verifier. Break down a complex management consultant, Re:Work Consulting,
question into a set of simpler ones, essentially Kazakhstan
building blocks that ladder up to the big
question. This can make a question that feels
impossible to answer more approachable.

• Fact checklist. As project professionals,

we want to make fact-based decisions.
This pattern allows practitioners to build a
fact base with identified sources, which can
provide a solid basis on which to make a
decision.

• Reflection. Sometimes the AI’s responses

might seem like they lack context or fail
to make sense. This pattern asks the AI
to explain its thinking, making the tools’
responses more easily understood.

Few-shot and chain-of-thought prompting

techniques are very helpful with the
augmentation of project tasks.

22
4.0 AUTOMATING,ASSISTING AND AUGMENTING PROJECT WORK

SIDEBAR

Key Variables to Tailor When Using GenAI in Project

Management
Researchers and product managers working on PMI Infinity™, the new PMI-developed AI tool,
provided some key variables that can immediately impact the results a GenAI tool provides.
Project professionals can leverage these key variables when designing a prompt for a GenAI tool:

1. Level of knowledge: 2. Project life cycle or 3. Persona:

management approach:
A. Project management A. Project manager
knowledge: A. Predictive B. Team member
i. New to project B. Iterative C. Stakeholder in the company
management C. Incremental D. Stakeholder outside the
ii. Medium level of project D. Adaptive/agile company
management knowledge E. Hybrid
iii. Project management
expert

B. Industry:
i. New to the industry
ii. Medium level of industry
knowledge
iii. Industry expert

C. Project management
approach:
i. New to the approach
ii. Medium knowledge of
the approach
iii. Expert in the approach

23
5.0

Further Guidance
5.1 Excellent Prompting References
Prompt patterns and the prompting techniques discussed in this
report are certainly not the only ones out there. In fact, providers
of popular GenAI tools — including OpenAI,11 IBM,12 Microsoft13 and
Meta14 — have created their own prompt engineering guides that
enable users to learn how to create prompts that best fit with their
context.

24
5.0 FURTHER GUIDANCE

5.2 Continue Your PMIxAI Journey

The PMI Artificial Intelligence in Project Management hub is a comprehensive source for resources to
guide you on your AI journey. You’ll find:

• Generative AI Overview for Project Managers: This is an online, introductory PMI course on GenAI, that
is free for PMI members.

• Artificial Intelligence Online Community: Connect with colleagues to learn and share about AI.

• Thought Leadership: Curated content from PMI about how AI is impacting project management.

11OpenAI. (n.d.). Prompt Engineering. OpenAI.

12 IBM. (n.d.). What is prompt engineering? IBM.
13Maeda, J., & Bolanos, M. (2023, December 13). What are prompts? Microsoft Semantic Kernel. Microsoft Learn.
14 Facebook Research. (n.d.). Facebook Research/Llama Recipes/Examples. README. Meta.

25
6.0

Appendix:
Plug-and-Play
Prompt Ideas
for Project
Professionals
Relying on recognition, not recall, can help project professionals to yield the
best results. To that end, creating a library of plug-and-play prompts can
help you optimize a tool tailored to your specific use of GenAI.

A collection of plug-and-play prompts can be a valuable resource for

project teams working with GenAI models. Such a library can contain a set
of predefined prompts that are designed to elicit specific responses from
the AI model.

26
6.0 APPENDIX: PLUG-AND-PLAY PROMPT IDEAS FOR PROJECT PROFESSIONALS

This approach can help project teams to save time and effort when developing prompts,
and ensure the prompts are effective in guiding the LLM to produce the desired output at
the same time. Further, it can standardize the prompt engineering process across different
GenAI tools and team members.

The example prompts provided in this appendix are a solid starting point. Over time, you will
want to refine these to reflect your personal style and preferences. Note that they include
a robust description of a persona and a clear example of the intended output. You can add
other tasks that could benefit from automation, assistance or augmentation along the way.

Directions: Please replace the highlighted words below with your specific project details.

6.1 Example Automation Prompt: Meeting Minutes Summary

Play the role of an expert project manager with a deep understanding of [your industry]
and [the project type you are engaged in]. Please summarize the following meeting minutes.
The tone of the document should be practical, clear and credible.

Here is the meeting transcript.

[copy in transcript]

Instructions: Follow the seven-step process detailed below to summarize meeting minutes.
Follow each step completely and in order. The goal is to produce the highest quality
document possible.

“When encouraging your

project teams to use GenAI,
make sure you build a library
of sorts for your team.
They’re going to end up using
it a lot faster if they’re
going to select it from a
library, instead of coming up
with something from scratch.”
- Nathan Rischling, program manager, Sigma
Defense Systems, USA

27
6.0 APPENDIX: PLUG-AND-PLAY PROMPT IDEAS FOR PROJECT PROFESSIONALS

1. Begin with metadata: 6. Review for completeness and accuracy:

• Date of meeting: • Ensure all discussed topics are covered.

• Time: • Verify the correct assignment of action items
• Meeting title: and responsibilities.
• Participants: • Check for any grammatical or spelling errors.
• Location/platform:
Sample meeting minutes format:
2. Read the transcription:
Meeting title: [Title]
• Start by reading the entire meeting Date: [YYYY-MM-DD]
transcription carefully to understand the overall Time: [HH:MM]
flow and key points discussed. Participants: [Names]
Location/platform: [Location/platform]
3. Identify key sections:
Opening remarks:
• Opening remarks: Briefly summarize the - [Summary of opening remarks and objectives.]
opening remarks, including the meeting objective.
Discussion points:
• Discussion points: For each topic discussed, 1. [Topic 1]
note the key points, including any data or figures - [Key point 1]
mentioned. - [Key point 2]
2. [Topic 2]
• Decisions made: Clearly state any decisions - [Key point 1]
that were made, including any voting outcomes, if - [Key point 2]
applicable.
Decisions made:
• Action items: List all action items, specifying the - [Decision 1] - Agreed upon by [Names].
task, the person responsible and the deadline. - [Decision 2] - Decided after [Brief context].

• Issues raised: Note any issues or concerns that Action items:

were raised, including who mentioned them and - [Action 1]: Assigned to [Name], due by [YYYY-
any proposed solutions or follow-up actions. MM-DD].
- [Action 2]: Assigned to [Name], due by [YYYY-
• Closing remarks: Summarize any closing MM-DD].
statements or next steps mentioned toward the
meeting’s end. Issues raised:
- [Issue 1]: Mentioned by [Name], with [Proposed
4. Summarize statements by participants: solution/follow-up action].
- [Issue 2]: Highlighted by [Name], needs further
• For each section, briefly summarize what each discussion.
participant said, focusing on contributions to
decisions, action items or issues raised. Use bullet Closing remarks:
points for clarity and to keep it concise. - [Summary of closing remarks, including any
mention of the next meeting.]
5. Format for clarity:
7. Check for errors:
• Use bullet points and numbered lists for easy
reading. • Use a spelling and/or grammar tool to ensure
• Highlight action items and decisions in boldface accuracy.
for emphasis. • Reread the minutes to ensure they accurately
• Assign clear responsibilities for action items, reflect the meeting’s discussions and outcomes.
using names or initials.

28
6.0 APPENDIX: PLUG-AND-PLAY PROMPT IDEAS FOR PROJECT PROFESSIONALS

6.2 Example Assistance Prompt: Generating a Risk Matrix

Imagine you are an expert project manager in the [Industry], focusing on [Project type]
projects. You have encountered a range of challenges specific to this domain, including
[List specific challenges]. Your task is to develop a comprehensive risk matrix that addresses
these unique challenges, providing a clear overview of potential risks that might impact
project success.

For each identified risk, include the following details:

• Risk name: A brief title that encapsulates the essence of the risk.

• Description: A detailed explanation of the risk, including how it specifically relates to the
[Project type] in the [Industry] and the identified challenges.

• Level of impact: Assess the potential impact of the risk on the project’s objectives,
categorizing it as high, medium or low. Consider factors such as cost overruns, delays and
quality issues.

• Likelihood: Evaluate the probability of the risk occurring, categorizing it as high, medium or
low, based on your expertise and experience in the industry.

The matrix should cover a variety of risk categories, including but not limited to budgetary
concerns, scheduling conflicts, technological challenges, regulatory changes, resource
availability, scope creep, stakeholder engagement, data security, quality assurance and
external dependencies.

The goal is to create a visually engaging and informative risk matrix that not only highlights
potential risks, but also serves as a strategic tool for mitigating these risks effectively. Your
matrix should enable project stakeholders to quickly understand the risk landscape and make
informed decisions to ensure project success.

The format of the risk matrix should be in a table that can be copied and pasted into any
other document. Mirror the example below.

29
 Level of Impact
Likelihood (High/
Risk Name Description (High/Medium/
Medium/Low)
Low)

Budget overrun Cost exceeds budgeted High Medium

amount due to
unforeseen expenses or
underestimation of initial
budget.

Schedule delay Project timelines extend High High

beyond initial estimates
due to various factors
such as resource
availability or scope
changes.

Technology risks Implementation of new or Medium Medium

untested technology could
lead to integration and
functionality issues.

Regulatory changes Changes in laws or High Low

regulations during the
project life cycle could
impact project delivery or
costs.

Resource availability Difficulty in securing Medium High

necessary team members
or materials can delay
project milestones.

Scope creep Uncontrolled changes or High High

continuous growth in the
project scope without
adjustments to time, cost
and resources.

Stakeholder engagement Lack of stakeholder Medium Medium

support or engagement
can lead to misalignment
with project goals and
requirements.

Data security Risks associated with High Medium

data breaches or loss
of sensitive information
during the project.

Quality issues The final deliverables High Medium

may not meet the
required standards or
specifications, leading to
rework.

External dependencies Reliance on third parties Medium Medium

or external entities that
could affect project
timelines or deliverables.

30
6.0 APPENDIX: PLUG-AND-PLAY PROMPT IDEAS FOR PROJECT PROFESSIONALS

6.3 Example Augmentation Prompt: Framing a Decision

Imagine you are an expert project manager in the [Industry], focusing on a [Project type]
project. You have encountered a complex question, which is “[Complex question?].” Your task
is to break this complex question down into smaller questions that collectively answer the
complex question, providing a clear view of the component parts of the problem.

Consider:
• Range of outcomes implied by the complex question.
• Sources of resistance.
• Potential risks and mitigation plans.
• Key project milestones and their contributions to overall objectives.

Instructions:
• Consider and compare alternative solutions or tools in terms of benefits, costs and risks.
• Define criteria for the final decision, ensuring alignment with organizational values and
priorities.
• Establish measures for decision success and feedback mechanisms for ongoing
improvement.

Provide one-to-three-sentence answers to each question developed. Then ask me where I

would like to go deeper.

31
7.0

Acknowledgments
• Bart Gerardi, Senior Manager, Software Engineering, Redfin,
United States

• John Signo, Vice President, VP DMT Delivery Lead, Bank of America, USA
United States

• Nate Rischling, Program Manager, Sigma Defense Systems,

United States

• Subhankar Modak, PMO Program Manager, Accenture, India

India

• Pamela Krzypkowska, Director of Research and Innovation, Ministry of Digital Affairs,

Poland

• Olalla García Pérez, Deputy Project Manager, Kawasaki Rail Car,

Spain

• Yuliya Zhevno, Project and Change Management Consultant Re:Work Consulting,

Kazakhstan

32
7.0 ACKNOWLEDGMENTS

About PMI Thought Leadership

We are a multidisciplinary team of subject matter experts, experienced leaders and
researchers dedicated to creating, retaining and disseminating innovative and thought-
provoking project management research and content. Partnering with the PMI community,
industry thought leaders and prominent authorities, we build and sustain a community-
generated knowledge platform. Thought Leadership is dedicated to offering cutting-edge
perspectives, contributing meaningfully to solving problems, and providing solutions to a
broad, diverse global community spanning from early career professionals to executives.
We are committed to the diversity of opinions and community contributors to represent all
voices of the project management profession equally.

Our work has three core areas of focus:

• Individuals – Equip project professionals with high quality, actionable recommendations to

drive individual growth, performance and continuous learning.

• State of the Profession – Inform project professionals of the latest trends and practices
to improve project management effectiveness and success as well as to advance the
profession.

• Enterprise & Innovation – Inspire and provide strategic direction to senior executive leaders
to help transform their organizations for long-term growth.
Grounded in analytical insights and practical recommendations, our work empowers our
community to successfully navigate dynamic business landscapes and society.

About Project Management Institute (PMI)

PMI is the leading authority in project management, committed to advancing the project management
profession to positively impact project success. We empower professionals to excel in project
management practices through our growing global community, knowledge sharing, and best-in-class
certifications—driving positive change in organizations and communities. Since 1969, our unwavering
mission has been to advocate for the profession by offering life-long learning and connections to
sharpen high-demand skills. Today, PMI provides professionals at every stage of their career journey
with the globally recognized standards, online courses, thought leadership, events, and tools they need
to succeed. With more than 300 chapters around the world, PMI members can network, find mentors,
access career opportunities, and learn from peers, working together to drive greater impact.

Visit us at www.pmi.org, linkedin.com/company/projectmanagementinstitute, facebook.com/PMInstitute,

and on X @PMInstitute.

33
 PMI.org

©2024 Project Management Institute. All rights reserved. “PMI” and the PMI logo are marks of Project Management Institute, Inc.
For a comprehensive list of PMI trademarks, contact the PMI Legal Department.