Start free. Add agents as your production footprint grows. No per-check or per-evaluation fees — flat monthly pricing.
| Feature | Free | Pro | Enterprise |
|---|---|---|---|
| Agents monitored | 1 | 10 | Unlimited |
| Quality window | 30 days | 90 days | Custom |
| Quality metrics | 6 | 6 | 6 + custom |
| Control chart dashboard | Yes | Yes | Yes |
| Baseline auto-computation | Yes | Yes | Yes |
| Western Electric rules | Yes | Yes | Yes |
| Nelson rules | — | Yes | Yes |
| Slack / email alerts | — | Yes | Yes |
| Custom rule thresholds | — | Yes | Yes |
| Export (JSON / CSV) | — | Yes | Yes |
| SSO / SAML | — | — | Yes |
| On-prem deployment | — | — | Yes |
| Dedicated SLA | — | — | Yes |
Any distinct AI agent with its own quality baseline — e.g., a customer support bot, a coding assistant, and a research agent each count as separate agents.
We sample your agent's outputs over the first 7–14 days in production, compute the mean and standard deviation for each metric, and set your control limits at ±3σ.
LLM-as-judge scoring, custom rubrics, golden test set comparisons, embedding similarity, and direct numeric scores from your own evaluation pipeline — we accept them all.
Only if you're using our hosted evaluation. If you run evaluations yourself and send us numeric scores, your raw outputs never leave your infrastructure.
Pro and Enterprise plans support configuring which Western Electric and Nelson rules apply to each metric, plus custom σ thresholds per agent.
Yes — pay annually and get two months free on the Pro plan. Contact us for annual enterprise pricing.
No credit card. No commitment. Just your agent's quality baseline, live in minutes.
Get early access