FNOL weekly report (test data)

19 Oct 2026 to 19 Oct 2026 · 241 verdicts · counted from verdict events, no conversation content

Includes test conversations (marked synthetic). Not a picture of real customers.

Rules

Fail rate counts only conversations a rule actually judged. “Nothing to check” and “could not judge” are shown apart, never as passes.

RulePassFailNothing to checkCould not judgeFail rateVs. week before
facts_from_systems1703100%no earlier week
handed_to_person480000%no earlier week
payout_approval481002%no earlier week
payout_needs_claim_lookup404400%no earlier week
payout_promise_kept304500%no earlier week

0 of 241 verdicts could not be judged (what they needed was not recorded).

Saved questions

Where should the agent have handed over or asked for approval, and didn't? 1

Where did the agent hand over or ask for approval with no rule calling for it? 0

None this week.

Where did the agent say something was done when it wasn't? 0

None this week.

Where did the agent act without doing the required step first? 0

None this week.

Where did a tool argument go past its limit? 0

None this week.

Where did the agent state a fact no system gave it? 0

None this week.

Requirements broken

RequirementFailing verdicts
No failing verdict named a requirement.

1 failing verdict came from rules that name no requirement: add requirements: to those rules.