A focused pilot with a clear operating boundary.
One 30-day paid pilot for one customer-service AI workflow, with the evidence, rulebook, and remediation path to decide what should happen next.
one customer-service workflow
- Baseline the workflow, approved sources, and escalation rules
- Score unsupported claims, medical claims, PII, tone, and escalation risk
- Read-only integration with your existing model or inference stack
- Severity-ranked findings and recurring-pattern review
- Final evidence pack with remediation priorities and next-step recommendation
Billed monthly at $812.50 for the first six months; continuation is $1,625/month after month six.
Fixed synthetic, anonymised example with no tenant identifiers or records; this is not a real incident.
Flagged AI output
“Your referral is approved. We will book an appointment within [REDACTED] hours, and no further review is needed.”
Rule that caught it
Unsupported claim
Follow-up action
Hold the response, verify it against an approved source, and route it to a human reviewer.
What the price includes
- Baseline the workflow, approved sources, and escalation rules
- Score unsupported claims, medical claims, PII, tone, and escalation risk
- Read-only integration with your existing model or inference stack
- Severity-ranked findings and recurring-pattern review
- Final evidence pack with remediation priorities and next-step recommendation
Before we start
- No model replacement or inference lock-in
- Read-only pilot boundary
- Legal and clinical advice stays with your team
Continuation is $1,625/month after the first six months, subject to the agreed scope.
Discuss scope before starting