💡

InsightHunt

Hunt the Insights

M

Mike Krieger

Episode #213

Chief Product Officer (CPO) at Anthropic

Anthropic

⚡Execution🎯Product Strategy

📝Full Transcript

13,935 words
Lenny Rachitsky (00:00:00): 90% of your code roughly is written by AI now. Mike Krieger (00:00:03): The team that works in the most futuristic way is the Claude Code team. They're using Claude Code to build Claude Code in a very self-improving kind of way. We really rapidly became bottlenecked on other things like our merge queue. We had to completely re-architect it because so much more code was being written and so many more pull requests were being submitted. Over half of our pull requests are Claude Code generated. Probably at this point it's probably over 70% that it just completely blew out the expectations of it. Lenny Rachitsky (00:00:26): You guys are at the edge of where things are heading. Mike Krieger (00:00:28): I had the very bizarre experience of I had two tabs open. It was AI 2027, and my product strategy, and it was this moment where I'm like, "Wait, am I the character in the story?" Lenny Rachitsky (00:00:36): It feels like ChatGPT is just winning in consumer mind share. How does that inform the way you think about product, strategy, and mission? Mike Krieger (00:00:43): I think there's room for several generationally important companies to be built in AI right now. How do we figure out what we want to be when we grow up versus what we currently aren't or wish that we were or see other players in the space being? Lenny Rachitsky (00:00:55): What's something that you've changed your mind about what AI is capable of and where AI is heading? Mike Krieger (00:01:01): I had this notion coming in like, "Yes, these models are great, but are they able to have an independent opinion?" And it's actually really flipped for me only in the last month. Lenny Rachitsky (00:01:12): Today, my guest is Mike Krieger. Mike is chief product officer at Anthropic, the company behind Claude. He's also the co-founder of Instagram. He's one of my most favorite product builders and thinkers. He's also now leading product at one of the most important companies in the ...

💡 Key Takeaways

  • 190% of code at Anthropic is written by AI, shifting bottlenecks from engineering implementation to decision-making and merge queues.
  • 2Product managers should focus less on specs and more on strategy, comprehensibility, and 'opening eyes' to what's possible.
  • 3The 'Make the Other Mistake' prompting technique: If the model is too nice, explicitly ask it to be brutal or roast your ideas.
  • 4Successful AI products require the convergence of three elements: Model Intelligence, Context/Memory (MCP), and Application UI.
  • 5Don't build features that just create dependency; build for user agency and augmentation.
  • 6Founders can survive by focusing on deep vertical workflows (e.g., legal, biotech) or differentiated go-to-market strategies.

📚Methodologies (3)

⚡ Execution

When AI writes 90% of the code, the traditional PM-Designer-Engineer handover breaks down. The bottleneck shifts from 'writing code' to 'decision making' (upstream) and 'merge queues' (downstream). Teams must re-architect their infrastructure and review processes to handle this velocity.

Core Principles

  • 1.Shift Prototyping Left: PMs and designers use tools like Claude Artifacts to build functional prototypes, not just mockups.
  • 2.Re-architect Critical Paths: Optimize merge queues and CI/CD pipelines as the volume of Pull Requests (PRs) explodes.
  • 3.AI-Assisted Review: Use AI agents to review AI-generated code, focusing human effort on acceptance testing rather than line-by-line syntax checks.

"We really rapidly became bottlenecked on other things like our merge queue... Over half of our pull requests are Claude Code generated. Probably at this point it's probably over 70%... or 90%."

#ai-native#development#execution
View Deep Dive →
🎯 Product Strategy

To get high-quality strategic critique from an AI, you must explicitly push it to the opposite extreme of its training. If it's too nice, ask it to be brutal. If it's too shallow, force it to 'think hard' about reasoning before answering.

Core Principles

  • 1.Roast the Strategy: Explicitly instruct the AI to be brutal, critical, or 'roast' the user's ideas to break its politeness filter.
  • 2.Request Reasoning: Ask the model to 'think hard' or output its reasoning chain before giving the final answer.
  • 3.Meta-Prompting: Use the model to write its own system prompts (using tools like Prompt Improver) because AI understands its own XML tag structures better than humans do.

"With Claude sometimes I'm like, 'Be brutal, Claude, roast me. Tell me what's wrong with this strategy.'... make the other mistake."

#'make#other#mistake'
View Deep Dive →
🎯 Product Strategy

Useful AI products are not just about the model. They require a convergence of three distinct layers: Model Intelligence, Context/Memory, and Application/UI. Focusing only on one leaves the product incomplete.

Core Principles

  • 1.Model Intelligence: The raw capability and reasoning power (Research team's output).
  • 2.Context & Memory: The bridge connecting the model to proprietary data (solved via MCP - Model Context Protocol). Without this, answers are generic.
  • 3.Application & UI: The workflow layer that makes integrations discoverable and results actionable.

"For utility of AI products, it's three part. One is model intelligence, the second part is context and memory, and the third part is applications and UI."

#utility#equation#strategy
View Deep Dive →
← Browse All Episodes