How it works
Score every call. Let QA listen only where it matters.
Call QA assists your QA team: every recording is scored against your checklist, then reviewers jump to flagged moments instead of listening from zero.
- 01
Ingest recordings
Submit a single recording URL, upload a CSV/Excel batch export from your dialer, or connect a source system. Duplicate recordings already processed that day are skipped automatically so you don’t burn cycles twice.
- 02
Transcribe & diarize
Workers run GPU-accelerated speech-to-text and separate Agent vs Customer speakers. Warm transfers to a closer are handled so licensed-agent introductions and recorded-line disclosures aren’t mislabeled as customer violations.
- 03
Score against your rules
Each campaign carries a free-text QA checklist and a separate zero-tolerance red-flag list. The model scores criteria 0–10 with exact evidence quotes. Any observed red flag forces the call score to zero — independent of how well the rest of the call went.
- 04
Lookup DNC / TCPA / blacklist status
Call QA queries compliance list APIs (DNCScrub, Blacklist Alliance, TCPA Litigator List, and related federal/state DNC sources) and your custom datasets portal API — so number status sits next to the scored call for QA and export.
- 05
Assign via round robin
Scored leads are distributed to QA agents on the campaign by round robin. Managers can reassign or change ownership whenever coverage or priority changes.
- 06
Human review with audit trail
QA agents work their assigned queue — filter to flagged or low-score calls, jump to evidence, then accept or reject with a comment. Human decisions stay separate from the model score.
- 07
Manager reporting & export
Managers see agent progress and quality performance (who’s strong, who needs coaching). Export the filtered set as CSV (Excel-ready) with AI QA notes, agent comments, and review status.
What reviewers see
Criterion scores, red flags, and the exact transcript line that produced them — with one-click jump into the recording.
Transcript
Final Expense · Lead #48291
Agent · 00:08
This call may be recorded for quality and training. I’m a licensed agent calling about final expense coverage.
Customer · 00:22
Okay… I’m 67. I’ve had some heart issues in the past.
Evidence · DNC red flag
02:14“Please put me on the Do Not Call list. I don’t want any more of these calls.”
Jump to moment in recording →
Agent · 02:21
Understood — I’ll note that and end the call.
QA scorecard
0 / 10
Zero-tolerance flag forced score to zero
- 9
Opening disclosures
Recorded line + licensed agent
- 10
Age qualification
Confirmed 50–85 range
- 0
DNC handling
Customer requested Do Not Call
- 8
Health screening
Disqualifiers reviewed
Human review
Tracked separately from the model score
Deployed per organization
Call QA is configured against your own campaign scripts and compliance requirements — not a one-size scorecard. Processing scales with your backlog via an autoscaled GPU worker fleet so capacity matches real dialer volume.
- Per-campaign YAML checklists and agent scripts
- Role-based access for admins, managers, and QA agents
- Optional email OTP two-factor authentication
Ready to speed up your QA floor?
Pilot Call QA with your QA team. We’ll configure your checklist and red flags, score a real batch, and show reviewers how to work from flagged calls with evidence.