Tiaan van Antwerpen

Writing

Where the thinking goes once it has settled.

Constraints, measurement, routing, and the gap between a model that scores well and a system someone trusts on a deadline.

  1. Oct 2026 · 5 min readThe factory moved into the inboxWhy building agentic AI is an industrial engineering problem, and what a hike in The Goal has to do with it.Read it
  2. Dec 2025 · 2 min readThe model is not the hard partNotes from forward-deployed enterprise AI work: the model is one component among many, and rarely the one that decides whether a project succeeds.Read it
  3. Nov 2025 · 6 min readAdaptive LLM routing for effective agentic workflowsMy MEng research: routing a text-to-SQL agent across thirteen models. The RAG router landed on the Pareto front at a fifty-sixth of the cost of Claude Opus 4, and on correctness alone, none of it beat simply picking a good model.Read it