🤖
AI Quality Auditor
AI-powered call auditing tool to empower agents.
Built on Firebase • Gemini • AssemblyAI
How it works
Four simple steps
Compass handles the heavy lifting. You focus on coaching and quality.
1
Upload call
Add MP3 + metadata (university, domain, type).
2
Transcribe
Diarized transcript with timestamps.
3
Compute metrics
Talk time, gaps, holds, overlaps.
4
Score & coach
Rubric scores with evidence + notes.
Scorecard v2.0
Fourteen Key Behavioral Criteria
Scoring Guide: Criteria are weighted to a total of 120.
Any point deductions must be supported by evidence.
Any point deductions must be supported by evidence.
Criteria & weights
- C1 Counselor Credibility & Role Positioning — 8
- C2 Call Responsiveness & Opening Presence — 8
- C3 Student Identification & Context Awareness — 8
- C4 Purpose Clarity & Call Objective — 15
- C5 Discovery & Needs Assessment — 8
- C6 Active Listening & Conversational Control — 8
- C7 Student Connection & Engagement — 8
- C8 Process Transparency & Call Control — 7
- C9 Communication Quality — 8
- C10 Language Quality & Professional Expression — 8
- C11 Information Accuracy — 10
- C12 Ethical Communication & Boundaries — 7
- C13 Closure Quality & Next-Step Ownership — 7
- C14 Overall Call Impression — 8
Evidence-first scoring
Each deduction references a timestamped snippet with a short coach note.
FAQ
You might be wondering…
Short, honest answers. Replace with your specifics.
Does this work with Next.js?
Yes. It’s a React component and works with Next, Vite, CRA, etc.
Where do transcripts live?
In Firestore/Storage collections tied to each call; access is role‑restricted.
How is scoring explained?
Every deduction includes a timestamped snippet and a short coach note.
Can we export reports?
Yes—export summaries and details to Google Sheets for sharing and audits.