Notes from the workbench
I build AI systems that have to work in the real world.
I’m Gareth Austen, a software and machine-learning engineer in Dublin. This is where I write about optimising Agents, developing robust quality gates, designing good evals, building Tell Me Ted, working with coding agents, and turning prototypes into dependable products.
Latest writing
All writing →-
Codex Can Write the Code. You Still Own the System.
Codex accelerated a major refactor, but an essential OTA capability needed repair. Here is how we changed our prompts, tests and review process.
-
Building a Startup With ChatGPT as a Daily Collaborator
A practical account of how a small team uses ChatGPT across product thinking, software, hardware, operations and decision-making and where it still fails.
-
Building a Teddy Bear That Can Read With a Child
Tell Me Ted is a teddy bear designed to read alongside a child. Here is why we started building it, how the prototype works, and what we are learning.
Series in progress
Building Tell Me Ted
Twelve field notes on building a physical AI product: the architecture, the hardware, the judgement calls and the mistakes.