Prompts & analysis

Questions the AI checks on every call. Adjust them per client, then re-run the analysis on a chosen period.

Green6 of 6 active

Pitch adherence: these questions produce the agent score (0–100%).

standard · g1

standard · g2

standard · g3

standard · g4

standard · g5

standard · g6

Orange5 of 5 active

Mid-severity violations: a Telegram notification goes to management.

standard · o1

standard · o2

standard · o3

standard · o4

standard · o5

Red4 of 4 active

Critical violations (privacy, fraud): instant notification + mandatory review.

standard · r1

standard · r2

standard · r3

standard · r4

Re-run analysis

apply the current question set to already transcribed calls

Pick a period and press “Run”. The demo runs on mock data; in production this is the Claude API over stored transcripts.