Research
We build owned AI products, so most of what we learn comes from shipping them, not from a whiteboard. This is where we will write it down: how AI agents behave in production, how to make RAG trustworthy on a business's own data, and what it really takes to replace rented SaaS with software you own. The first writeups are in progress.
The patterns that survive real data, real load, and real users, and the ones that quietly fall over.
How we ground our products in a business's own data without the answers drifting or leaking.
What it takes to move off per-seat SaaS onto software a team owns, source, data, and infrastructure.
