Thinking on AI
in Manufacturing
From operators who build production systems — not people who write about them from the outside.
The Values Ceiling
We gave one AI six companies to run, each with its own values. On the decisions that mattered, it kept overriding them and running itself. The research behind why enterprise agents need their own tuned models.
Why AI Productivity Is Not Reaching Revenue or Margin
Companies running AI report more work getting done. Revenue and margin have not moved. The gap is a resource allocation problem, and every old rule about scarcity still applies.
The Future of Small, Tuned Models in Manufacturing
Why the most useful AI in your plant won't be a giant general-purpose model — it'll be a small one trained on your operation.
Why Agentic Workflows Are the Real Unlock for Manufacturing
Chatbots answer questions. Copilots assist individuals. Agentic workflows actually run processes. Here's why the distinction matters for your operation.
The Hard Problem Isn't Storage — It's Retrieval
Every company has the data. The real challenge is getting the right information to the right decision at the right time.