Jason (Xingjie) Liu

🍜Anybody up for some ramen? I am excited to share that I am back in the garage, building again (it’s actually a house in the Berkeley Hills). We are taking the wraps off of Supercog AI, a new startup focused on solving Application Integration using LLM powered agents. My co-founder in this new venture is a great friend of mine (and a former co-founder partner), Alex Osborne. Alex and I have worked around the application integration space for a long time, and we are unrealistically excited about applying the GenAI stack to this problem. Not many people relish the task to get some bits to move from system A to system B. But it’s a critical job to be done, and that’s led to a lot of brittle scripts, a myriad of inflexible “no code” tools, and a lot of hand crafted SQL. But the power of the Large Language Model, trained on a huge corpus of information, offers a remarkable tool to solve this problem. The LLM you use today already knows the APIs of hundreds of popular systems. It knows the SQL dialects of every major database. It understands the semantics and data schemas of many popular SaaS systems. It understands much of the specific domain in which your business operates. I like to call the AI revolution “the last platform shift”. That may turn out to be hyperbole, but I definitely believe we are in the very early stages of seeing what this new stack can enable. I know we are hardly the first folks to claim that “amazing stuff is coming!”. But this is why Alex and I have decided to focus on a real and hard problem. If we can prove that LLMs can power a new way to solve this problem, it will be the first successful *new* approach in 15 years.. We will have much more to share about this new platform soon. In the meantime, if you’re interested in getting a peek - or even better if you have application or data integration tasks that you would like some help with - please reach out because we want to help. #genai #startups #backinthegarage

163

37 Comments

Organizations have insufficient technical capabilities for the speed and scale of their ML model implementations. My colleague Sumit Agarwal just published a very helpful research with a capability view and sample vendors for defining a modular, flexible and scalable MLOps solution. https://lnkd.in/gnB_T9bq #MLOps #LLMOps #AI

149

17 Comments

𝑳𝑳𝑴-𝒂𝒔-𝑱𝒖𝒅𝒈𝒆 🧑‍⚖️ 𝘓𝘓𝘔 𝘌𝘷𝘢𝘭𝘴 & 𝘵𝘩𝘦 𝘪𝘯𝘴𝘪𝘥𝘦 𝘴𝘵𝘰𝘳𝘺 𝘰𝘧 𝘩𝘰𝘸 Twilio Segment 𝘣𝘶𝘪𝘭𝘵 𝘊𝘋𝘗 𝘊𝘰𝘱𝘪𝘭𝘰𝘵 𝘵𝘰 𝘢𝘤𝘤𝘦𝘭𝘦𝘳𝘢𝘵𝘦 𝘮𝘢𝘳𝘬𝘦𝘵er 𝘸𝘰𝘳𝘬𝘧𝘭𝘰𝘸𝘴 3𝘹 These days I'm spending an irrational amount of time thinking about evals. It started with a conversation months ago with Jaimal Soni, when he was showing me an absolutely 🔥🔥 demo of what he's been building in stealth. I couldn't believe what I'd just seen. "What's one of the hardest problem you had to solve?" I asked. "Evals." Evals are the way to assess (ahem evaluate) LLMs and systems built with LLMs. In software engineering, you'll hear a lot about test coverage and unit tests — little tests to verify that your code is doing what you expect. But generative AI & LLMs are a little different: unlike your deterministic code, models are stochastic, meaning you never really know what the output will be until it's generated!Ask chatGPT the *same* question and you will get *different* answers. This is where LLMs get their magic, but also creates problems when you're asking an LLM to solve discrete tasks. As we've been building out CustomerAI & Generative Audiences, we needed to think about this problem from the start. Generative Audiences allows marketers to write natural language to generate new audiences, without having to navigate a sophisticated audience-building UI. We’ve been blown away by the results of – customers are experiencing a 3x improvement in median time-to-audience creation, and world-class feature retention rate IF the audience generation works on the first attempt. So evals are 🔑 Ankit Awasthi, Salman Ahmed, Alfredo Láinez Rodrigo, & James Zhu came up with a really clever solution to ensure that CDP Copilot is generating audiences that are accurate & represent the intent of the prompt. But it wasn't easy...and today we're sharing how we used synthetic data & LLM-as-Judge architecture to accelerate marketer workflows 3x.

90

8 Comments

I participated in a “SMART MANUFACTURING” panel with Gayathri Radhakrishnan, Rohini Chakravarthy and Abhijit Supekar (moderator) last week at TIECon 2024. The coolest part of the panel was 3 Women VCs talking about a topic that has not traditionally been discussed by women. I never imagined I’d see this in my lifetime. Things have indeed changed! We had a very lively discussion of whether we can expect lights out factory in 5-10 years. Automation is happening in all stages but full automation manufacturing is tough because it requires human like adaptability. It is not converting text into instruction or code - it is about precision. Robotics + AI is making this happen and we will get there. AI is going to bring the cost down for complete solution for example: there are robotic arms that can do more tasks and learn like a human and not just some specific tasks. Also check out our portfolio companies Micropsi Industries and RGo Robotics that are making it happen in this space. Thanks to Anshu Bahadur Hrishikesh Sathawane Nilesh Dixit for organizing the panel and Anita Manwani and Surbhi Kaul for making this happen! #TiECon2024 #AIUbiquity

95

6 Comments

Founder Dadagiri: The challenge with Founders who "just can't let go". I keep sharing Microsoft's incredible story. I saw Steve Ballmer and met Satya Nadella, and they were as different as Neebu ka achaar and Boondi ka Ladoos. (Ballmer was Achaar). But they (Ballmer and Gates) let Nadella do his thing; boy, what he did was magical! Enter Howard Schultz, the founder and genius of Starbucks. Via WSJ “Howard Schultz, the legendary entrepreneur who turned a local coffee chain into a global icon, was meeting Laxman Narasimhan, the man who would succeed him as Starbucks CEO in a month. Shultz had recently watched a New York Giants practice and was inspired by how football teams manage their time and structure their work by unit—offence, defence, and special teams. He told Laxman that the same principle could be applied to Starbucks’s systems, which Schultz had spent nearly a year analysing in search of ways to improve service. Narasimhan wasn’t buying it. “That’s a football team. We are a company,” he said. “If you don’t like it, that's fine; it’s your company,” Schultz said, laughing, as Narasimhan chuckled.” It seems that the spirit is not working. "It all exploded into public view Sunday night when Schultz wrote a LinkedIn post that read like an open letter to shareholders criticizing how Narasimhan and his senior leaders were running the business" See link - https://lnkd.in/dx3ebFgP I read the post, and these statements sounded terrible. See #Dhandhekibaat (DKB) comments inserted in the flow "There is a natural tendency to try to do too much too soon. " "Don’t try to do everything at once." "Leaders must model both humility and confidence as they work to restore trust and increase performance across the organization." DKB - It sounds like he is preaching to some B'School kids. The Humility and Confidence point sounds horrible "Over the past five days, I have been asked by people inside and outside the company for my thoughts on what should be done. The stores require a maniacal focus on the customer experience through the eyes of a merchant. The answer does not lie in data but in the stores." DKB - therein lies the bullying. WHY answer these questions and give side views? Instead, say, "I am sure Laxman and the team are on top of it" (And send your gyaan privately to Lax). Learning: As a founder, it's impossible to let go. Only the rarest manage to achieve Microsoft's magic. As they say, "Love means never having to say sorry". #Dhandhekibaat "Letting go means never having to say ANYTHING" :) Do you agree? Article Link in comments

85

8 Comments

🌟 Unlock the Power of Generative AI with Andrew Ng! 🌟 DeepLearning.AI has launched "Generative AI for Everyone"—a groundbreaking course designed to make generative AI accessible to all. 🚀 🎓 What You'll Learn: Core Concepts: Understand the fundamentals of generative AI and how it works. Real-World Applications: Discover how tools like ChatGPT, DALL·E, and more are transforming industries. Business Strategies: Learn how to integrate generative AI into your organization to boost productivity and creativity. Practical Insights: Hands-on exercises to help you grasp the technology, even without a technical background! 👩‍💻 No coding experience? No problem! This course is tailored for everyone—professionals, business leaders, and anyone curious about AI's potential. ✨ Ready to shape the future with generative AI? Enroll now and start your journey today! ➡️ deeplearning.ai #GenerativeAI #ArtificialIntelligence #Upskill #AIInnovation

5

A new research paper by our incredible teams at Google Research and Google DeepMind shows the strong performance of Med-Gemini across a variety of multimodal tasks, including  2D and 3D imaging in radiology, histopathology, ophthalmology, and dermatology, and genomics. Med-Gemini ranked best in class on chest X-ray and CT scan report generation, and is able to perform genomics-based disease prediction and mortality risk more accurately than a standard linear polygenic risk score (PRS) based approach. This work couldn't have been possible without close collaboration with our partners, including a panel of specialists who assessed and rated our models as part of this research. These promising results showcase the potential of LLMs to transform medicine as we know it. https://lnkd.in/dHx8_DYh

289

5 Comments

#Rise of Micro LLMs: A Roadmap to Business Success with Generative AI In the current AI landscape, Large Language Models (LLMs) like GPT-3, BERT, and T5 have garnered significant attention for their impressive performance on various NLP tasks, including text generation, question answering, summarization, and translation. However, the training and deployment of these massive, general-purpose models come at a substantial cost, often reaching up to $10 million. These expenses cover not only computational resources but also specialized hardware such as GPUs and the necessary software infrastructure. Micro LLMs are developed by refining larger models to address industry-specific, domain-specific, and customer-specific scenarios, offering several key advantages: #Personalization: Enhances user engagement and satisfaction by providing more relevant interactions, leading to higher retention rates. #Efficiency: Requires less computational power and memory, making them suitable for deployment on smaller devices or in edge computing scenarios. #Accuracy: Reduces hallucinations and bias by focusing on enterprise-relevant data. #Cost-effectiveness: Eliminates the need for expensive data scientists or costly hardware and systems. While general-purpose or "macro" LLMs are beneficial in scenarios requiring a broad understanding, their wider scope may not always align with the specific nuances of enterprise-focused tasks. Additionally, their flexibility and adaptability can sometimes come at the expense of precision compared to Micro LLMs. #Benefits and #Limitations Micro LLMs offer significant benefits, such as being trainable on modest datasets and having simplified architectures that enhance interpretability. Their compact size makes them ideal for deployment on mobile devices, and their ability to process data locally is particularly valuable for IoT edge devices and organizations with stringent privacy and security needs. However, their limitations include more restricted knowledge bases due to smaller training datasets and a less comprehensive understanding of language and context, which can lead to less accurate and nuanced responses compared to larger models. #The Future of #Micro LLMs The future of small language models looks promising with increased accessibility and cost efficiency, making them attractive for businesses and individuals. Their lower computational requirements enable real-time processing on edge devices, enhancing privacy and energy efficiency. These models are easier to fine-tune for specific tasks, ensuring effective and specialized solutions. Additionally, their faster response times and seamless integration with existing systems improve user experience. As advancements continue, small language models will become more capable, promoting the democratization of AI and fostering innovation across various fields. #GenerativeAI #Micro-LLMs

20

Please join me on June 25th San Francisco, CA, USA for a panel titled "A BLUEPRINT FOR SCALABLE & RELIABLE ENTERPRISE AI/ML SYSTEMS" w/ moderation by Hira Dangol from Bank of America and my fellow panelists Rama Akkiraju from Nvidia and Nitin Aggarwal from Google. Recent breakthroughs in AI technology have captured the world's attention, and now it's time to address the challenge of creating truly robust and reliable AI systems. AIQCon's mission is to harness the power of our community to develop new GOLD STANDARDS for AI quality. The global movement is bringing together renowned leaders and professionals from across the industry to establish reliable AI quality solutions and standards. If you're working on AI systems, you won't want to miss this one! Use the discount code "Community" for a 15% discount on registration. Register at https://lnkd.in/gqRE48fC. #AI #EnterpriseAI #AIQuality #AIQCon https://lnkd.in/gqRE48fC

58

3 Comments

Excited to share my first blog post in a new series: "LLM Study Diary: A Comprehensive Review of LangChain — Part 1" After a brief hiatus of 6 months from LangChain, I found the landscape had changed significantly. This inspired me to revisit Greg Kamradt's excellent tutorial, but with a twist: In this post, I: - Walk through Kamradt's example codes - Update and correct parts that don't work with the latest libraries - Share insights and solutions for compatibility issues It's a practical guide for anyone looking to use LangChain with current versions. Read more: https://lnkd.in/gtbmetkH #AI #LangChain #LLM #Python #CodeTutorial Have you encountered similar update challenges? Share your experiences!

9

I'm excited to finally share what Sunder Karthik and I have been working on for the past four months: Lintbook, The Cash Flow App. Here is a 90 second demo - https://lnkd.in/gB7prnVs Why did we build Lintbook? 💬 Navigating Tough Money Conversations Talking about money, especially in a relationship, isn’t easy. Personally, I lean toward saving for the future, while my partner prefers to focus on the present. Both perspectives matter, but finding a financial balance requires ongoing, honest communication. Lintbook was born from our own journey of trying to bridge those gaps, and it’s designed to help you do the same. 📈 Focusing on What Matters, Not Rigid Budgets One of the biggest reasons we created Lintbook was to move away from rigid, joyless budgeting. Budgets can feel like crash diets - they’re hard to stick to and often make you feel restricted. Instead, Lintbook focuses on tracking cash flow in a way that fits your life, without sacrificing the things that matter to you. 🔧 Built from the Ground Up with Intention When we first came up with the idea for Lintbook, we didn't want to create just another finance app - it was about crafting a tool that we’d want to use ourselves. Instead of packing it with features that can overwhelm users, like automatic bank syncing or endless charts, we kept it simple and intentional. The idea was to give people more control over their finances, without the extra noise. So why did we take this back-to-basics approach? For us, it’s like “flintstoning” - going back to the way our parents did it, manually entering expenses into a ledger. That process is grounding. It puts you in control and invites conversations when you’re ready, without the flood of automated data overwhelming you. Spending just a few minutes tracking our income and expenses has transformed our relationship with money, reducing anxiety and helping us focus on how to grow our cash flow. We hope Lintbook brings you that same sense of peace. Key Features: 📸 Itemized Receipts: Snap a picture of your shopping receipt, and we’ll itemize it for you. (Special shoutout to Peter Marsh for helping build this) 🔄 Automated Recurring Transactions: Plan your regular bills or income. ⏰ Never Miss a Bill: Get reminders for upcoming and overdue payments. 📊 Sub-Transaction Tracking: Break down large expenses for better clarity. 💡 Insights: Identify large expenses and view aggregates such as yearly taxes paid and category based spend 📅 Plan Ahead: Add future transactions to avoid surprises. 🤝 Collaborative Money Management: Work together with your partner at your own pace. 📝 Manual Ledger with a Modern Twist: Track your expenses with intention, just like our parents did. Available exclusively for iPhones in the US 🇺🇸 🔗 Download Lintbook https://lnkd.in/gv6Vm-mP 🌐 Learn more: www.lintbook.com Thank you for your support! I can’t wait to hear how Lintbook helps you on your financial journey!

101

18 Comments

If you think compute scaling will plateau, think again. We are in the baby states of inference time compute. VLLMs coupled with computer tools, reward learning and RL can leverage orders of magnitude more inference time compute than what we have seen with foundation models. Here’s a concrete training architecture (replace \pi with a large pre-trained vLLM and add it to the reward learner too, \R is the space of all computer tools)

21

1 Comment

Krish Naik, a fantastic AI mentor on YouTube, and his latest video about Chat LLM by Abacus.AI. For those interested in Generative AI, this video is a goldmine! Krish dives deep into Chat LLM, a powerful tool that allows you to: ● Use a variety of large language models (LLMs): Choose from open-source models like Llama 3 or paid options from OpenAI and Claude 3. ● Analyze code and generate responses: Get help with coding tasks and have the model interpret and respond to your code. ● Work with attachments: Upload PDFs or images and have Chat LLM answer questions based on their content, or even analyze images! ● Collaborate with a team: Invite team members and work together on projects. ● Connect to external services: Integrate Chat LLM with services like Slack, Teams, and Google Drive for a seamless workflow. ● Build custom chatbots: Train your own chatbot using your data to answer specific questions. And the best part? Chat LLM is significantly cheaper than competitors like Chat GPT, with a monthly subscription of only $10! In the video, Krish walks you through all these features with clear demonstrations. He even creates a custom chatbot based on a PDF document, showcasing its capabilities. If you're looking for a versatile and affordable LLM platform, highly recommend checking out Krish's video and exploring Chat LLM! Link to Krish Naik's video: https://lnkd.in/gYrcgMsE Let me know in the comments if you have any questions about Generative AI or Chat LLM. hashtag #genai hashtag #llm hashtag #opensource hashtag #chatllm hashtag #krishnaik hashtag #ai hashtag #machinelearning

3

1 Comment

Worth reading: Generative AI is not going to build your engineering team for you. So, so good - peppered with gem quotes and observations- chuckle. https://lnkd.in/gZBKtU4X Thanks to Eric Brooke for sharing.

16

What a great talk from Swapnil Sayan Saha, on Physics-Aware Tiny Machine Learning! https://lnkd.in/g9g4YWs4 There's exciting times ahead as Physics-aware TinyML is becoming increasingly important in merging the power of machine learning with operating in the physical world. By integrating real-world physics models into ultra-low-power devices, this approach dramatically enhances the accuracy and efficiency of TinyML applications. From improving energy efficiency in smart sensors to enabling more accurate anomaly detection in industrial applications, this approach to machine learning at the edge helps machines better understand their environment - without sacrificing power or performance. As we explore the possibilities, this combination of physics and Edge AI opens up new opportunities in fields like healthcare, environmental monitoring, and autonomous systems. It's a huge leap toward making AI more sustainable and impactful, and allows us to build more useful AI-enabled systems that operate in (and on!) the physical world. #TinyML #EdgeAI #PhysicsAwareAI #AIInnovation #Sustainability #MachineLearning

4

So much activity in this area. I need an AI tool just to keep up ;-)

5

Breaking News: HANOOMAN. AI Shatters Boundaries, Pioneering a New Era of Global AI Accessibility In a major step forward for technology, India introduces Hanooman.ai, propelling the country to the forefront of global innovation. Named after the revered figure from Hindu Epic, Hanooman.ai showcases India’s tech skills and its aim for a more inclusive digital future. Hanooman.ai is incredible—it can talk in 12 Indian languages and over 90 international languages. This means it can connect people from different cultures and places like never before, marking a big moment in AI development worldwide. But Hanooman.ai isn't just about languages. It also helps communities by letting them talk in their own languages online, making sure everyone's voice is heard. This is a big deal and shows how technology can bring us all together. And there's more! Hanooman.ai is just the beginning. Soon, it'll do even more—it'll understand speech, audio, video, and images. This will open up lots of new possibilities, from helping people with disabilities to changing how we do customer service. Hanooman.ai is changing the game for AI. It's making sure everyone can use advanced technology, no matter where they're from. Whether it's connecting people who speak different languages or sparking new ideas, Hanooman.ai is making a difference. If you want to see what Hanooman.ai can do, you can find it as an app on the Play Store or use it on the web. Whether you want to chat in different languages, try out cool AI features, or just be a part of something big, Hanooman.ai is ready for you. In short, Hanooman.ai shows India's commitment to using technology for everyone's benefit. As we celebrate this big moment, let's look forward to all the amazing things Hanooman.ai will bring to the world of AI. Congratulations Vishnu Vardhan and entire team of HANOOMAN. AI #HanoomanAI #AIInnovation #GlobalAccessibility #LinguisticVersatility #TechRevolution #InclusiveAI #DigitalEmpowerment #MultimodalAI #TechForAll #FutureOfAI

108

7 Comments

At booxar, we wanted to solve the issue of LLMs hallucinating when people interacted with books. Integration with LlamaIndex.TS helped us address this problem. Now we are able to provide contextual responses to our readers by leveraging a Retrieval-Augmented Generation (RAG) system. It ensures our LLMs have up-to-date, semantically relevant context, which improves the accuracy of our Question and Answer systems and chatbots. LlamaIndex is helping us ingest, structure, and access contextually relevant data. With LlamaIndex.TS, we're enhancing the reading experience by making interactions with books more reliable and informative Check it out at https://beta.booxar.com #booxar #LlamaIndex #RAG #AI #ReadingExperience #ContextualResponses

9