Connect
JustCall (call recordings) connects with an API key and secret pasted from your JustCall dashboard. Intercom (Conversations & Tickets) connects via OAuth — no key to copy, just approve access.
How it works
From a closed call or ticket to a coached agent
Ingest
Calls are transcribed and speaker-diarized. Tickets — whether uploaded as a PDF or pulled live from Intercom — are normalized into the exact same shape before scoring, so the engine can’t tell the two sources apart.
Score
Every agent who touched the case is evaluated independently against every rubric dimension. Every verdict cites the exact line it’s based on.
Review
Team Performance surfaces the trend and the weak spots. Every number is clickable straight back to the real evidence behind it.
Coach
Training turns each agent’s own real Top Gap findings into Listen / Drill / Recap exercises.
Why evidence citation matters
PASS
A verdict on Resolution Correctness doesn’t just say Pass — it shows the exact line (“I’ve processed yours as a one-time exception”) a manager can check against the real thread in one click.
Why per-agent scoring matters
GAP
A ticket handed off mid-thread. The first agent’s diagnosis was solid; the second agent’s handoff dropped context and made the customer repeat themselves. One shared score would hide which agent actually needs the coaching — two independent scorecards don’t.
Honest answers
Can a ticket be rescored?
Off by default — Claude isn’t perfectly deterministic, so a re-run could change a score that’s already been acted on. An org can opt in.
What if we change our rubric mid-quarter?
Rubrics are versioned — you can build a new one, activate it going forward, and old scores stay tied to the rubric that produced them.
Is our data isolated from other customers?
Yes — two independent layers, described on the Product page’s trust block.
See your own calls and tickets scored — with the receipts.
Request a demo