billiem

Home / topic collection

AI coding

Personal tools and open-ended experiments built with coding agents, from a stock tracker to a native Mac dictation app. The posts show what the agents produced, how the results were checked and where they failed.

Start here

  1. A refurbished office chair stock tracker that saved me over £100

    A concrete starting point: a temporary stock tracker, with a real purchase and useful checks beyond whether the code ran.

  2. LLM Choice: using autonomous coding agents to find project ideas

    The method behind the open-ended experiments: let agents choose a project, then judge what they actually made.

  3. Prompting GPT-5.6 for deeper coding experiments

    How a more demanding brief changed the experiments, and why reviewing their output still mattered.

  4. Building and testing a local Mac dictation app with Codex

    A larger tool with an installed app, repeatable comparisons and microphone testing that exposed failures after repository checks passed.

More writing

Latest first.

Your CLAUDE.md is not magic

Why I want AI-written code checked independently, with shared standards and room for developers to choose their own workflows.

What GPT-6.1 Sol chose to build

Three autonomous GPT-6.1 Sol experiments: a fictional record, a sheet-cutting planner and a congestion model, with their review fixes and limits.

What GPT-6 Astra chose to build

What three Astra agents built with room to choose: a string puzzle, a supply-chain experiment and a printable booklet, with their fixes and limits.