Production Readiness Score
A composite score appears only when enough verified evidence exists across quality, reliability, approval, evidence confidence, latency and revision burden.
Weights
Quality 30% · Reliability 25% · Approval 15% · Evidence 15% · Latency 10% · Revision burden 5%
Minimum evidence gate
3 verified quality scores · 5 reliability runs · 3 cost observations · 3 latency observations · at least 1 active evidence assertion.
Missing buckets make the score insufficient_evidence; missing data is never converted into zero.
Provider readiness
Why this matters
Enterprise teams need more than a beautiful demo. A production-ready provider should repeatedly meet quality targets, clear approval efficiently, respond quickly and be supported by current evidence. VoicePilot keeps every underlying component inspectable so a composite never hides missing data.