A practical guide to when to use ChatGPT, Claude, and agent modes
Simon Willison · rss · 2026-07-28
Simon Willison highlights how Ethan Mollick’s recommendations have shifted from chat-first usage to agentic systems that can do hours of work in one go.
Key takeaways:
- Gemini has fallen off Mollick’s list because Google still lacks a clear ChatGPT Work / Claude Cowork-style entry point.
- The most powerful pattern is giving AI access to your computer through ChatGPT or Claude apps.
- ChatGPT’s modes are Work and Codex; Claude’s are Cowork and Code, and the naming is confusing.
- On mobile, ChatGPT Work is especially surprising: switching from Chat to Work removes the usual internet restriction in the Code Interpreter container.
More from coding & agent
- Nativ v0.1.0 Released: Run AI Models Locally on Apple Silicon with Multi-IDE Support — max_paperclips · 2026-07-28
- CodeGraph made local token counts spike, but provider logs showed the run used fewer tokens — TruthTellerTom · 2026-07-28
- Locally Hosted Kimi K3 Beats Cloud GPT-5.6 in 3D Physics Generation — rohanpaul_ai · 2026-07-28
- Hermes Desktop Avatar wraps a local agent in an always-on-top chat UI — TinfoilTricorn · 2026-07-28
- Berkeley-led paper maps three autonomy levels for software development — dawnsongtweets · 2026-07-28
- The real risk in AI coding is teams skipping the verification level — dawnsongtweets · 2026-07-28