How Anthropic teams use Claude Code

Author: Anthropic · Type: practitioner · Status: draft · Published: 2025-07-24 · URL: https://claude.com/blog/how-anthropic-teams-use-claude-code

Licence: Copyright Anthropic; free to read, quote with attribution.

Summary

Internal case studies of ten Anthropic teams operating with Claude Code: autonomous-loop patterns with human review before merge, test-driven development with periodic check-ins, and non-technical staff (legal, marketing, data science) shipping working tools. The consistent pattern is supervised autonomy - let the agent run, review before consequences - rather than unattended operation.

Insights

• Anthropic’s own internal norm is supervised autonomy: agents work in autonomous loops but humans review solutions before final refinement and merge - a template for a solo founder’s engineering pipeline. (high, draft) human in the loop autonomy agents in production

  They give Claude abstract problems, let it work autonomously, then review solutions before final refinements. — Product Design and Security Engineering team sections

• Agentic coding lets one person cover functions that previously required specialists, which is the mechanism by which a solo founder plausibly covers engineering, content, and ops simultaneously. (high, draft) solo founder agents in production case study

  Agentic coding isn’t just accelerating traditional development. It’s dissolving the boundary between technical and non-technical work, turning anyone who can describe a problem into someone who can build a solution. — Non-technical teams section

• The highest-yield mental model is agent as thought partner embedded in human workflows, not as an unattended replacement for them. (medium, draft) human in the loop orchestration

  The most successful teams treat Claude Code as a thought partner rather than a code generator. — Cross-team lessons