Pilot · $0
Why we'd take your team on Pilot
Not a spray waitlist. Pilot is for beachhead teams — Teams that already have an agent or workflow on one labeled, high-volume job — stuck because a general or frontier LLM is still flaky, expensive, or unsafe with proprietary data even with loops and tools around it.
- FitYou already have the agentHarness in flight. We don't sell another agent — we replace the weak LLM call for one scoped job.
- FitYou have labels (or can get them)Historical outcomes for fraud, triage, routing, or review. Without labels we can't measure beat-frontier.
- FitFrontier is the bottleneckToo expensive at volume, too flaky on the loop, or too risky with proprietary context leaving your perimeter.
What you get
Shared cloud. Enough to eval + train one scoped task model.
Measure on your held-out outcomes before you swap the agent call.
Rate-limited task-model endpoint your harness can call.
How to ask (copy for outreach)
“You're running an agent on [job] where the frontier LLM is [cost / flaky / data-path]. We'd put a task model under that call — small open weights on your outcomes — and prove it on your eval in a free Pilot. Why you: [agent + labels + volume].”
Start Pilot
Train a task model on Shared. Upgrade to BYOC when data can't leave.