ProductsProduct / feature launchSpeech & Audio1 source · Oct 9, 2026

A new feature for my blog, built using my voice

I used the ChatGPT desktop app for this, in the Codex tab, using the voice conversation mode, running against a local development environment.

Proof1 independent outlet

Key points

  • I built the feature almost entirely using my voice, chatting away to my laptop while I cooked dinner.
  • I started the session against my local simonwillisonblog checkout by typing:
  • I had a pretty good idea of what I wanted to build, and it's a simple enough Django feature that I was certain the model (in this case GPT-6 Astra High) would be able to do it.
  • Here's an extract of my voice transcript that was captured by Codex:

Sources (1)

  • [1]A new feature for my blog, built using my voice
    Simon Willison's Weblog · Oct 9, 12:54 PM
    I used the ChatGPT desktop app for this, in the Codex tab, using the voice conversation mode, running against a local development environment.
    I built the feature almost entirely using my voice, chatting away to my laptop while I cooked dinner.

Extractive summary: sentences quoted from the sources.

Hype check

Independently coveredCovered by 1 outlets

  1. Follow-up · Oct 9, 2026ICYMI: What landed for AI builders in September 2026
  2. Follow-up · Oct 9, 2026Can machines reason non-verbally?
  3. Follow-up · Oct 9, 2026Can Language Models Play the Wikipedia Game?
  4. Follow-up · Oct 9, 2026Microsoft's Decision-1 model enters the fast-growing AI decision model race
  5. Follow-up · Oct 9, 2026pydantic/pydantic-ai v2.55.0: v2.55.0 (2026-10-09)
  6. Follow-up · Oct 9, 2026On the Depth Transfer of Neural Optimizers via L-Smoothness

Before this

  1. Oct 9, 2026Asana cuts model costs 76x in browser tests with GPT-6.1 Sol
  2. Oct 8, 2026[AINews] not much happened today
  3. Oct 8, 2026How Oracle turns days of work into minutes with ChatGPT and Codex
  4. Oct 8, 2026A Closer Look at Agentic BBO: Benchmarking LLM Agents for Black-Box Optimization
  5. Oct 8, 2026SWE-Duel: Towards a Self-Scaling Arena for Coding Agents
  6. Oct 7, 2026[AINews] Claude Haiku 5.5 — better than GPT-6 Luna at the same pricing

Related