Skip to main content

Unified Agent Quality Scorecard (AI + Human)

Make the manual human Agent Scorecard also score AI agents — automatically — on one CRM-native quality lens, and surface it for QA. The initiative consumes the SkillPack engine's 9-metric evaluator (the AI judge already ships; and the scorecard already auto-scores the human agent via auto_agent_scoring.rb — but nothing scores the AI agent), fuses it with the existing human QA scorecard via a two-tier rubric (9 Qontak-calibrated defaults in a separate group + org-owned custom params using the existing prompt field), and scores per actor/segment so a room handled first by an AI agent then a human is scored for both. This is Move 3 / Gap G2 of the Qontak AI Agent strategy.

QA Lane

Lane B — keeps a human QA gate. No Lane-B trigger is present (post-hoc analytics/scoring), but no E2E test specs exist for this initiative yet, so the Lane-A entry bar (100% E2E, spec-mapped coverage) is unmet — Lane B by default. Classified 2026-06-29.

Master index (ANCHOR)

See the ANCHOR for the full phase index, north-star metrics, and initiative-level decisions. There is no Confluence ANCHOR page for this initiative; the per-phase Confluence PRDs are linked in the table below (frontmatter confluence: points to the lead Phase 1 page).

PhasePRDConfluenceStatus
Phase 1 — Scorecard Settings & Rubric Configprds/phase-1-settings-and-rubric-config.mdQON/51229163544📝 Draft
Phase 2 — AI Auto-Scoring & In-Room Scorecardprds/phase-2-auto-scoring-and-in-room-scorecard.mdQON/51228868664📝 Draft
Phase 3 — Unified Analytics Reportprds/phase-3-unified-analytics-report.mdQON/51228770359📝 Draft
Phase 4 — Validation / Testing Harness— (shared with ai-agent-testing)⏳ Not started
Phase 5 — Go-Live Gate⏳ Not started
Phase 6 — Self-Improvement Loop⏳ Not started
Phase 7 — Calibration & Ecosystem⏳ Not started
Phase 8 — Multi-Agent Scoring & Selectable Scorecard (parked)prds/phase-8-multi-agent-scoring-and-selectable-scorecard.mdQON/51229229057📝 Draft