# How Anthropic Builds: Lessons from Labs — Mike Krieger, Anthropic

## Executive summary

Mike Krieger discusses the evolution of building with AI models, emphasizing that ambition requires being 'unreasonable' in tool usage and process design. He details advanced development workflows—such as porting large codebases (e.g., Python to TypeScript) over a weekend using LLMs—and outlines modern organizational structures for research labs (like Anthropic's), which prioritize rapid iteration through 'persevere or pivot' cycles rather than rigid project management. Key technical shifts include moving from task delegation to expressing end-state goals, and utilizing advanced features like tagging for multi-agent, proactive code development.

## Key takeaways

- Be Unreasonable in AI Usage: The first generation of AI products often constrained access to tools and degrees of freedom. To maximize potential, users must be 'unreasonable' in their prompts, moving beyond simple task delegation to defining the desired end state.
- The Importance of Intent Over Code Lines: In large codebases (e.g., 2,000 line PRs), the bottleneck is no longer review time but human comprehension. The trend is shifting toward sharing 'Claude code artifacts' that explicitly lay out intent, trade-offs, and rationale alongside the code.
- Labs Structure for Rapid Iteration: Anthropic Labs operates on a two-week 'persevere or pivot' cycle. Teams are assembled around 'bets,' and the structure deliberately avoids tying people to specific projects, allowing for quick disbandment and reassembly.
- Scaling Requires Pre-Measurement: Drawing from Instagram's scaling experience, it is crucial to pre-measure everything that might be needed (metrics) and implement thoughtful knobs and feature flags for dynamic runtime configuration.

## Technical details

- Code Migration/Porting: Successfully porting a large, legacy codebase (e.g., hundreds of thousands of lines from Python to TypeScript) and verifying it using dynamic workflows over a short period.
- AI Development Workflow Shift: Moving from writing detailed strategy documents requiring AI critique (task delegation) to describing the desired end state and allowing the model to iterate through trade-offs autonomously.
- Multiplayer Delegation via Tagging: Advanced usage of LLMs involves 'tagging' agents, making them responsible for monitoring specific code sections or feedback channels and proactively taking on tasks (e.g., fixing bugs, updating APIs) rather than just responding to direct prompts.
- Data Provenance in Finance AI: In financial services, the ideal model output must balance free-form analysis with strict verifiability and audit logging (data provenance) to meet industry requirements.

## Practical implications

- Adopt a mindset of 'unreasonable' ambition when designing AI workflows; focus on the end state rather than step-by-step instructions.
- Implement robust CI/CD practices that include pre-measuring all potential failure points and utilizing dynamic configuration (knobs/feature flags) for runtime control.
- Structure development teams in a 'bet' model with clear, short cycles ('persevere or pivot') to maximize rapid learning and minimize sunk costs on failing projects.
- Shift code review communication from line-by-line inspection to structured artifacts that document intent, trade-offs, and rationale.

## Topics

AI Development, Software Architecture, LLM Workflow, Build Engineering, Product Strategy, Scaling, Claude, Fable, Cloud Design / Cloud Code

Source: https://www.youtube.com/watch?v=qqrk7CtkuIw