Pricing · Transparent ladder
Pay for confidence, not commodity throughput
Enterprise sales logic: higher-confidence launches, fewer bad outputs, auditable human oversight, lower internal ops burden — not “cheaper labels.”
Step 1
Evaluation Pilot
$2,500 one-time
Narrow, high-value task family: domain-specific evaluation, red teaming, or exception-queue review.
- Rubric design workshop
- Qualified reviewer pool for the pilot
- Gold-set / benchmark setup
- Exportable audit pack
- Findings readout (async or call)
Step 2
AI Evaluation Ops
$3,500 / month
Recurring evaluation or approval capacity when work is continuous rather than one-off projects.
- Weekly delivery cadence
- SLA-aware routing
- QA summary dashboard export
- Adjudication included on contested items
- Priority expert escalation
Step 3
Human Judgment API
$499 / month + usage
Programmatic access to reviewer and expert queues once you embed human review in your stack.
- Task submit + result retrieve
- Webhooks & provenance metadata
- Skill-tier routing
- Sandbox keys for integration
What we deliberately defer
Open marketplace take-rates
Asset-light marketplaces need liquidity and trust first. We prove quality via services before marketplace economics.
Data licensing
High ticket only when provenance and rights are clean. Not a year-one distraction.
Custom enterprise SOWs available for multi-domain eval programs, regulated operators, and embedded agent oversight. Public ladder prices are starting anchors — pilots scope the real unit economics.
Data Day