Simon Willison
@simonw
AI should help us produce better code https://t.co/q8N9nJyLDk
30retweets
View on XSimon Willison's Guide Argues Coding Agents Can Raise Code Quality
Simon Willison· Updated
Shipping worse code with AI agents is a choice. Simon Willison's Agentic Engineering Patterns guide shows how teams can use agents to tackle refactoring debt — API redesigns, concept renames, module splits — that never felt worth the manual effort.
Simon Willison, creator of Datasette, argues in his Agentic Engineering Patterns guide that AI agents are ideal for refactoring tasks — conceptually simple but time-consuming fixes that compound as technical debt. API redesigns across dozens of files, renaming a core concept, splitting a large module — agents handle these via Gemini Jules, OpenAI Codex, or Claude Code on the web, surfacing results as a pull request to review, land, or discard.
Agents can also run exploratory prototypes before architectural decisions. A single prompt can wire up a load test simulation, validating a technology choice with evidence — multiple experiments run in parallel at near-zero cost.
The guide introduces the compound engineering loop: end each project with a retrospective documenting what worked, so agent instructions improve over time. Refactoring debt that was too costly to justify manually is now within reach — the cost has dropped to near zero.
Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →
