The evaluation habit that changes how teams ship.
Why a small, steady set of examples often beats a giant benchmark.
Read the noteThe Auralis journal
Field notes on product quality, model operations, and the work of making intelligent systems useful.
Why a small, steady set of examples often beats a giant benchmark.
Read the noteContext is more than a prompt window. It is the history around the decision.
Read the noteA short note on taste, restraint, and the teams we want to build with.
Read the noteThe design decisions that help people understand when to rely on an AI result.
Read the noteHow source quality, freshness, and permission shape the experience people receive.
Read the noteWhy we care about calm software, direct language, and work that can be revisited.
Read the noteA short letter, occasionally
One thoughtful note about building AI in production, every few weeks. No noise, no growth hacks.
Get new field notes in your inbox.