Blog
Enterprise AI, from pilot to production.
Field notes on getting AI past the demo and into production — written by the engineers who build and run it.
AI Pilot Purgatory: Why Enterprise AI Stalls Before Production — and the Path Through It
I've seen this pattern more times than I can count. The demo works. The numbers look good. Then nothing reaches production. Here's why pilots stall — and what the organisations that escape purgatory are doing differently.
Engineering
A Confident Wrong Answer Is Worse Than No Answer
Enterprise AI assistants rarely die in a crash. They die when a fluent, confident answer turns out to be wrong, trust erodes, and usage quietly fades. The fix isn't a bigger model. It's grounding. Here's what that means in production.
Governance
The CISO Said No: What Enterprise AI Governance Actually Means in Production
Most enterprise AI pilots don't fail on the model. They fail the security review. Here's what governance actually means in production: who can ask, what the system saw, and whether you can replay any decision a year later.
Enterprise AI
Build, Buy, or Partner: The Enterprise AI Decision, Made Honestly
Every enterprise weighs the same three options for AI: build it, extend a chat tool, or reuse RPA. Each stalls in a predictable way. Here's an honest read on all three, a fourth option, and the one question that decides between them.
See it in production
Production AI at named European enterprises — see the deployments.