The AI Runtime Lab

Skip to content

The AI Runtime Lab

Public engineering investigations for production AI systems. Not demos. Not tutorials. Measured attempts to make AI workflows reliable under real constraints.

Investigations

Investigation 01 investigating

Does selective source-verification buy enough precision to justify its cost and recall hit?

Customer

Two-person pre-seed dev-tools startup, no growth hire.

Updated

2026-06-25
- agents
- evaluation
- reliability
- retrieval