I use AI to produce architecture deliverables faster, and I help architecture teams do the same. One rule: nothing ships without human review.
What I bring to a team
- tools that automate documentation work: export in the client’s branding, diagrams, data extraction from an architecture repository, option evaluation through to the decision paper, critical review by a panel of expert profiles (CIO, CISO, DPO…);
- a review method: built-in quality gates, human in the loop;
- evaluation of AI use cases for IT, which I started as early as 2023 at a gas infrastructure operator.
See also Applied AI for architecture deliverables.
My tools
A library of about twenty skills and agents, reusable across projects and driven by configuration. The agents run in isolated containers, and every deliverable goes through quality gates before I review it.
GitHub Copilot and other assistants
I know the main AI coding assistants on the market, GitHub Copilot among them, and my library runs on several of them. Since 2026, at a car importer, I have been supporting the move of development tooling to GitHub and the rollout of GitHub Copilot Enterprise.
I sign what the tool writes
I use these agents in production on enterprise architecture deliverables. When a tool drafts the document, my job is to judge what it wrote and to answer for it. A generated document gets reviewed like a document you are about to sign, because that is what it is.