Vibeleaderboard
← All Intel
Intel / article

A new feature for my blog, built using my voice

Source
simonwillison.net
Date

Context

Simon Willison built a new Newsletters page for his blog mostly by talking to a coding while he cooked dinner, then finished it by typing. According to his post, he used the Codex tab of the ChatGPT desktop app in voice mode against a local copy of his Django site, with OpenAI's GPT-6 Astra High doing the coding. In about half an hour of conversation the agent produced a data model, import code for his Substack and GitHub newsletter sources, archive pages and search integration.

Voice was not enough for the last step. When he reviewed the pull request, one import ran Git in a subprocess, but one source lives in a private repository, so he asked for an API-based import and spent roughly another half hour typing prompts and adjusting the public pages before deploying. He writes that voice suits multitasking, while pasting error messages or highlighting code is more efficient for detail work, so he does not expect it to become his daily way of working.

Key takeaways · AI-distilled
  • Willison first typed "Start dev server and open in browser" in the Codex tab, then started a voice chat, so he could ask the agent to show new pages and follow its progress visually from across the kitchen.
  • About half an hour of talking to GPT-6 Astra High produced a Django model and migration, admin config, four working importers (RSS, an undocumented Substack API, two GitHub repos), archive pages and site search integration.
  • Code review in the GitHub PR caught one wrong choice: an import shelled out to Git in a subprocess, but pulling from a private repository made an API-based import the better fit, so Willison switched to typing to fix it.
  • Getting to a deployable PR took roughly another half hour of typed prompting. Willison says typing still wins for details, because pasting examples and error messages or highlighting code beats describing it aloud.
  • His verdict: voice coding suits multi-tasking more than daily use and would not work in a shared office. The live visual preview plus a keyboard fallback is what made it more capable than his earlier phone voice-mode sessions.
Terms in this piece · Glossary
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
Read the source simonwillison.net
Recommended reads
Comments

Checking sign-in…

Loading comments…