Eval-driven development: the only way we ship agents
We write the test suite before the agent. Here is how a golden set, a grader and a CI gate keep our AI systems honest.
Field notes
Things we learned
shipping AI.
Engineering notes, strategy and honest opinions from inside real builds. No hype, no listicles.
Eval-driven development: the only way we ship agents
We write the test suite before the agent. Here is how a golden set, a grader and a CI gate keep our AI systems honest.
When the answer is not an LLM (and how to tell)
About one in four strategy engagements ends with us recommending something more boring. Three questions we ask first.
On-device AI for mobile apps: what actually works in 2026
Speed, privacy and offline support make on-device models compelling. Here is where they fit — and where the cloud still wins.
Retrieval that holds up in production
Chunking, hybrid search and citations: the unglamorous details that decide whether a chatbot is trusted.
Shipping a full SaaS solo with AI-assisted development
What building AZ POS taught us about where AI coding tools speed you up — and where they quietly slow you down.
Keeping your cost per AI call honest
Model prices fall every quarter, yet bills keep rising. Four habits that keep AI features profitable.