Field notes on AI, data engineering and SaaS onboarding
What 13,000 Insurance Claim Documents Actually Look Like
96% were photographs of paper. The exact diagnosis was recoverable on one claim in ten. The whole book cost under $25 to read. What a real claims corpus teaches you about automating one.
Hugo MatthaeyWe Deleted the Approve Button, and the AI Editor Got Safer
A confirm prompt is not a feedback loop. It asks a human to predict the result of an edit they cannot see. Here is what replaced it, and the day we found out the agent had been working blind the whole time.
Hugo Matthaey
Your Users Can Now Connect Claude to Your Software Themselves
The Model Context Protocol turned your server into an OAuth resource server, and that one reclassification moved the setup work off the user. They paste a URL, sign in the way they already do, click Authorize. Here is the handshake, what it costs to build, and the five things that silently break it.
Hugo Matthaey
The Enterprise Security Questionnaire Is a Sales Document. Treat It Like One.
It arrives with an external scan already attached, and every yes commits you to producing a document. Here is what the auditor checks, the controls that actually move the answer, and what to fix before you reply.
Hugo Matthaey
Fractional CTO or Development Agency? A Decision Table
These two get compared because they cost about the same. They are not alternatives. One supplies decisions, the other supplies hands, and buying the wrong one is expensive in a way that takes about six months to become visible.
Hugo Matthaey
Custom Software Development: Cost, Steps, and a 2026 Guide
What custom software actually costs, how a project runs week by week, and the signs that custom is the wrong answer. With real price ranges instead of "it depends".
Hugo Matthaey
Cypress Testing: A Guide From a Team Running E2E in Production
Tutorials stop at the first passing test. The work starts after that: tests that fail one run in three for no reason, screenshots nobody compares, a suite that logs in fifty times. Here is what we learned industrialising all of it.
Hugo Matthaey
Have messy data slowing down your onboarding? Send us a sample. We'll show you what clean looks like.
