How Anthropic Uses Claude
summary
Anthropic runs Claude across Slack, Drive, email and Salesforce to let agents read context, create code and operate for hours autonomously, reporting 8x engineer productivity and that 80% of merged code is authored autonomously with about 20% of work only possible because of these tools. Using Claude Code, Co‑work, Claude Design/Tag and frontier models like Fable 5.1, teams run 8–9 hour agent sessions, auto‑generate PRs with grader risk scores, pregrade security alerts, and drop old RAG/prompt scaffolding as models absorb those capabilities.
AI-written summary · about
chapters (17)
- 0:00Anthropic's use of Claude
- 1:12Internal Claude use
- 2:17Autonomous coding evolution
- 3:40Model coherence drivers
- 5:45Productivity gains measured
- 7:29Five principles for Claude
- 8:10Public access to context
- 9:59Coding tools everywhere
- 10:43Interface tuning and models
- 11:46Claude in collaboration tools
- 13:33Next frontier of intelligence
- 14:35Failures for regression testing
- 16:01Outdated AI workflows
- 18:27Future model advancements
- 19:11Killing outdated projects
- 19:32Autonomous software case study
- 20:14Summary
related talks
16mBeyond the IDE: The Future of Developing Software With Agents
23mAnthropic
38mOne Android Dev, Ten AI Teammates: Scaling Delivery With Claude Code
30mBeyond the Hype: Using AI Coding Agents Today
01h 06mHow I Tamed Claude
44mClaude Code for Spring Developers
18mClaude Code & the evolution of agentic coding
24mClaude Plays Worth Stealing | Claude Code & AI Coding Tips