Notes from delivery
What we're learning building AI and software systems in production — written by the people doing the work.
Retrieval-augmented generation is the difference between an AI assistant that guesses and one you can put in front of customers. Here's how we design retrieval pipelines that hold up in production.
The teams that ship agents to production fastest aren't the ones with the fewest guardrails — they're the ones with the right-shaped ones.
A four-hour batch window sounds fast until you calculate what happens in the gap.
Internal developer platforms feel like premature optimization right up until the tenth team asks how to stand up a new service.
Most migration timelines fail for the same reason: they were built against the documented system, not the actual one.
The best AI governance frameworks are invisible to end users and load-bearing for everyone else.