Every call recorded automatically — voice and the agent's screen, time-synced, encrypted (AES-256), PCI auto-redacted, and transcribed speaker-by-speaker. Nothing falls through a sampling gap.
Voice Recording — 100% capture, dual-channel, encrypted, PCI auto-redact
Screen Recording — agent desktop, synced to the call timeline
Every call scored against your rubric and routed by confidence — the inbox sorts by what the AI is uncertain about, not by call ID. Each criterion shows its reasoning, the rule it cites, and the evidence line. One click to override; the override teaches the rubric.
Triage inbox — calls routed by AI confidence; review only what needs a human
Defensible scorecard — each criterion with its reasoning, cited rule and evidence
Quality and operational KPIs on every interaction, not a 5% sample — score distributions, pass rates, agent and cohort performance, and a builder for the cuts your team runs on.
Score & Ops — average score, pass rate and 12-week trends
People & cohorts — agent and team performance
Reports — operational & quality cuts, exportable to your BI
Turn scores into coaching and insight — each agent's weakest criterion with example calls, plus auto-discovered themes, sentiment trends, and which topics drag QA scores down.
Coaching loop — performance distribution and per-agent coaching
Per-criterion performance — which criteria drive deductions, ranked
Conversation Intelligence — themes, sentiment, topic → quality
Your team owns the rubric and the knowledge base and edits them in-app. Every change runs against an audit set and must clear an eval-gate before it publishes — versioned, with rollback.
Rubric & criteria editor — your scoring logic, in your hands
Knowledge base — the policy answers scoring is grounded in
Playbook settings — per-campaign configuration
