Anthropic Study: Domain Experts See Higher Success Rates with Claude Code
alex_verem · x · 2026-08-28
Anthropic released a study on Claude Code usage based on a privacy-preserving analysis of 400,000 sessions (235k users) over 7 months.
Key Findings:
- Division of Labor: Humans make 70% of planning decisions (what to do), while Claude makes 80% of execution decisions (how to do it).
- Expertise Premium: Users with greater domain expertise get more work done per instruction and achieve higher session success rates.
- Occupational Parity: On coding tasks, success rates across major occupations are nearly identical to software engineers, with verifiable outputs like passing tests or committed work.
- Usage Shift: Over 7 months, debugging time fell by nearly half, with usage shifting towards end-to-end agentic tasks (deploying, running code, analyzing data, writing docs).
- Value Increase: The estimated value of typical tasks rose by 25% on average, based on comparisons to freelance job postings.
More from coding & agent
- zai releases GLM-5.3 open-weight model for agentic coding and defense — zai-org · 2026-08-28
- Desktop AI agents track big long-term tasks using sprints — draginol · 2026-08-28
- Row-Bot 4.9 ships Buddy, an always-on-top desktop overlay for controlling agent runs — Acceptable-Object390 · 2026-08-28
- Three agent transcripts show scorers rejecting early fake-flag solves as non-causal — moyix · 2026-08-28
- LlamaIndex founder breaks down the long tail of enterprise PDF parsing — llama_index · 2026-08-28
- Finance MoE Ling-3.0-flash-Fin Free for a Month; Human Review Boundary Stands Out — Designer_Mouse_6109 · 2026-08-28