Field plays for AI optimization.

No theory. Each piece below solves a specific, recurring problem we see when enterprise AI bills get out of control — token spend, inference latency, vendor lock-in, or eval drift. Written by the people running the playbook on real engagements.

We publish here when we find something worth saying — usually after running it on a client's stack and seeing it move the P&L line. If you want the longer version, that's what the 1-Week AI Potential Evaluation is for.

Coming next

More field plays in the pipeline

On agentic loops that quietly burn budget · semantic caching when it works and when it doesn't · sovereign inference economics · and how to tell if your AI vendor is the bottleneck.

Soon

Want this run on your AI stack?

The 1-Week AI Potential Evaluation is the same forensic audit we use to write these pieces — applied to your token traces, your vendor contracts, your P&L.