Jeff Feng

Honored to have been invited to fal and Henry Shi’s Lean AI Leaderboard Celebration Dinner, celebrating some of the leanest and fastest-growing startups in SF. Here are some of the most overwhelming trends that were discussed among high-growth companies working with small, high-output teams: 1/ Headcount is no longer used as a success metric. As Henry Ward said, “Hiring means we failed to execute and need help.” Headcount is not a flex measure anymore. People aren’t asking “How big is your team?” anymore. They’re asking “How many customers do you have?” 2/ They're finding ways to leverage AI across every function. Every engineer uses Cursor or Windsurf. Every GTM leader is in Clay. The most efficient teams are built AI-first across the board. Thank you to Henry Shi, Burkay Gur, and Gorkem Yurtseven for putting this all together!

67

I took my second Waymo across Los Angeles last night, and the experience was largely excellent. From the smooth pickup to navigating the freeway (which is now supported here in LA!), the technology is undeniably impressive. However, the "last mile" wasn't quite perfect. I was dropped off on the left side of a two lane, one way street, directly in traffic rather than on the shoulder. Also, my destination was on the right. As I've written previously (https://lnkd.in/gHw7drrc), there is great promise in using technologies like AI in countless ways, from autonomous driving to talent measurement. However, my recent Waymo experience highlights two critical lessons: 1. Humans Must Own the Outcome: Whether you are a passenger in an autonomous vehicle or managing talent, you cannot delegate accountability to the machine. Technology is a powerful tool, but humans must own the final result. In hiring, as in transport, we must remain in the driver's seat to ensure the "drop off" (i.e. the final decision) is the right one. 2. The Feedback Loop is Non Negotiable: Without human feedback to these systems, technology will never improve. I made sure to provide feedback on my ride, and we must do the same in our talent systems. If we don’t provide constant, iterative feedback, we lose the opportunity to refine them into something better. Progress happens when human intuition meets technological scale. We provide the feedback; the tech provides the lift.

33

4 Comments

Had such a great chat with Vivien Lin! She brought so much wisdom and vivid insights to the conversation: 1. We're trying to work with our users, when they make money we make money. We try not to stand on the opposite. 2. From the start we decided to exit China market. We never touched US users. It's all because we are quite self-disciplined and self-regulated. 3. We consider whether this product or whether a feature can truly solve the problem for the user. 4. From 2018, 2019 is always put our top priority to think about how we bring more new user to the crypto industry. It's not like competing the existing user with other exchange. 5. Ideally, the final product should be you don't need to worry about what functions sit in what page. You just go to the AI or the AI assistant to tell him what you want. 6. The difference in the roadmap between crypto and traditional finance is rooted in how the industry got originated. For traditional finance, banking and payment that's naturally where finance comes from than lending than trading. At crypto, the very first need where the industry originated is mining. 7. I think, we are still in a very precious time window especially for exchange it to become a sort of mega platform in near future. 8. My suggestion is always be open minded - open minded to the technology, business model and even all those new ideas. 9. My suggestion to everyone, even if you are a legal, you are a product manager or you are in the operation team, is to study programming, learn how to code. 10. The slides always tell the good stories but you need to get down to the details to understand what's happening. 11. Before all those regulations put into place there has to be someone familiar with how possible the regulation is going to be implemented in the future and we can get prepared beforehand. 12. Without one major hack exchange cannot be a mature exchange. 13. If I know what they are doing I can read their code, I can get a very direct sense about the quality of the project instead of really looking at the powerpoint. Watch the full session here: https://lnkd.in/dGhbXdwr BingX

10

I just finished watching all of the top AI-related sessions from the World Economic Forum 2026, with leaders like Jensen Huang, Satya Nadella, and Dario Amodei. The consensus? The constraints that restricted our abilities to drive product innovations years ago have largely evaporated. Here are 5 "Impossible" Product Ideas from two years ago that are now engineering realities: 1. The "Self-Coding" Feature Suite 💻 • The 2026 Reality: The loop is closing. Anthropic predicts that within months, models will perform most maybe all of what software engineers do end-to-end. We are moving from "Human in the loop" to "Human in the lead," meaning products can now autonomously maintain and upgrade themselves. 2. The "Universal Shopper" That Actually Buys 🛒 • The 2026 Reality: "Agentic Commerce" is here. Visa has rolled out protocols where agents carry attribute payloads (spending limits, preferences) to buy natively on platforms. The buy button is moving from the merchant's site directly into the AI agent. 3. The "Physical" Health & Care Assistant 🏥 • The 2026 Reality: AI has moved to "Physical Intelligence." It now understands the laws of physics and biology. With humanoid robots slated for public sale by next year, the hardware finally exists to deploy the software we dreamed of. 4. The "Brute Force" Intelligence Engine 🧠 • The 2026 Reality: The cost of intelligence has collapsed. Inference costs dropped 99% from GPT-4 to GPT-5 mini. We can now afford to throw massive test time compute at problems, allowing the model to think for minutes before answering, drastically reducing hallucination. 5. The "Sovereign" Enterprise Brain 🔐 • The 2026 Reality: Sovereignty is the new standard. The focus has shifted to "Sovereignty of the Firm", distilling your own private data into a model you control, ensuring your IP doesn't leak into a generic API. The agents are ready to work. Time to find that dry-erase marker!

2

We teamed up with Stanford Institute for Human-Centered Artificial Intelligence (HAI) in their effort to train a fully open source Foundation Model. This means that not only the weights, but also the code, data, and processes are fully open sourced. To optimize efficiency and performance, the team combined the power of JAX/XLA and TPUs. Read all the gory details of the journey and the different choices made in our new blog post: https://lnkd.in/gTGdURZ5

187

9 Comments

This is what I was referring to as the “art of reliable” end of the new AI skill spectrum in my last post.

1

It is amazing how much of the world's product strategies exist to cater to one simple human behavior: we don't like dealing with friction. Pricing models assume customers won’t spend hours comparing options across ten marketplaces. Subscriptions assume most users won’t fight their way through intentionally complex cancellation flows. Auto-renewals assume people won’t track, review, and renegotiate every policy they hold. This has worked for the longest period because time, attention and patience are finite for us humans. But that reality is about to change given AI Agents have no concept of time or friction. Humans might not have the time to price compare across platforms, their AI Assistants do. Humans might not have the patience to go through 10 steps to cancel their subscriptions, their AI Assistants do. Humans might not know enough to negotiate better policy rates, their AI Assistants do. The principles that drove product design, growth strategies and experiences are likely to get challenged and this is already being seen with every single SaaS app in the market today. #CXAI #AIAssistants

34

3 Comments

Over the last few weeks, I’ve been running a series of live sessions for PMs, engineers, and AI-curious builders who want to go beyond buzzwords and actually reason about AI products. We focused on three themes: 🧩 1. Deconstructing RAG In this session, we broke RAG down into simple, practical pieces: - A clear retrieval + generation mental model for more reliable AI products  - Core building blocks: embeddings, chunking, vector stores, retrievers, and the end-to-end flow  - How to decide when RAG is the right answer – and when it isn’t  - The PM’s role in RAG quality: metrics, hallucinations, and experiments  - Real‑world constraints: latency, cost, data quality, context limits, and common pitfalls we see in the wild  🤖 2. Getting Started with AI Agents In this webinar, we covered: - What an AI agent really is, and how it differs from a traditional app or “just a chat UI”  - Agent architecture 101: tools, memory, reasoning loops, LLM orchestration - How to think about frameworks like LangChain - A step‑by‑step path to your first working agent: define use case → design → build → deploy  - Common pitfalls first‑time builders run into, and simple patterns that actually ship 🎯 3. Breaking into AI Product Management AI PM interviews today look very different from “generic PM” loops. - How AI PM interviews differ from traditional PM interviews  - The typical formats used by startups and larger tech companies  - The level of AI & GenAI understanding PMs are expected to demonstrate (without being ML researchers)  - How to structure answers across product sense, execution, and AI trade‑offs - Common mistakes candidates make – and what strong, nuanced answers look like We also ran live mock interview with participants and shared real‑time feedback on problem structuring, communication clarity, and AI/product depth. Watching these breakdowns gave people a “behind the scenes” view into how interviewers actually think. 🔭 What I’m seeing across all three Across RAG, agents, and AI PM interviews, a pattern keeps repeating: > People don’t just need more content on AI.  > They need better mental models and feedback on how they think about AI products. That’s the gap I’m trying to close through my work – combining AI literacy, product thinking, and career growth for PMs, aspiring PMs, and technical professionals.   --- 👥 What’s next – AI PM Cohort To go deeper into this, I’m also running an AI Product Management cohort for aspiring and current PMs who want structured practice, real product scenarios, and feedback. If you’re exploring a move into AI/GenAI product roles or want to strengthen your decision‑making as a PM, this cohort is designed to give you that combination of concepts, frameworks, and hands‑on application. Details on structure, schedule, and pricing are here: https://lnkd.in/gwMX-v-A

18

A product team we worked with was juggling multiple external vendors - development, design, QA - across different time zones. Deadlines were missed, communication was inconsistent, and the PMs spent more time chasing updates than building the product. We introduced an AI-powered vendor management tool that: - Monitors timelines and flags missed deadlines - Analyzes code quality and rework percentages - Tracks communication patterns, clarity, and collaboration - Measures response speed to ensure smooth coordination The outcome? - Managers had a single dashboard for all vendors - Problems were caught before they snowballed - Teams could focus on product delivery instead of firefighting The bigger picture: managing external teams isn’t just about keeping tabs, it’s about predicting issues and improving collaboration. At MintyMint, we use AI and data insights to help companies manage vendors seamlessly, reduce risks, and deliver faster. 👉 DM me if you’d like to explore it more. #MintyMint #AI #VendorManagement #Productivity #Collaboration #ProjectManagement #TechTools #DigitalTransformation

7

2 Comments