What Labs Actually Do With $100M Data Deals: Turning Company Workflows Into Agent Eval Sandboxes

StewartalsopIII · x · 2026-11-16

Everyone talks about labs spending hundreds of millions on data, but almost no one looks at what happens after the deal closes.

The core observation: researchers take company workflows and turn them into small sandboxes. Every tool call and edge case gets converted into an environment where agents can run thousands of trajectories until they stop breaking things.

The real value isn't the raw data — it's the decision logic hidden in the workflow, plus the built-in eval with outcome labels that comes with it.

Original post →

More from AGI Musings

AGI Musings channel →