Every engagement starts with a free pilot. Send us 5 tasks, in the exact format you'd give your own team. We'll send back expert-graded data within two weeks. No cost, no commitment.
The process
- Send us 5 tasks Use the format you'd give your own team: an environment spec, eval items, preference pairs, SFT prompts or red-team targets. Add your rubric if you have one.
- We match the experts We choose practitioners for your jurisdiction and domain from evaluators who have already passed our domain assessment, then calibrate them on your rubric.
- Expert-graded data within two weeks Every item comes back scored against your rubric, with written rationales and our agreement figures. You review it and tell us what to change.
- Scale when the pilot meets your bar Production work starts within 10 business days of a signed SOW, with the same experts and the agreement targets you set in the pilot.
What we need from you
- 5 tasks in your own format (RL environment, benchmark item, preference pair, SFT prompt or red-team target)
- Your rubric or grading notes, if you have them. Our experts can draft the rubric with you.
- The jurisdiction and domain the work covers, then the language
- The volume and timeline you expect after the pilot