Would We Catch It Now?
Replay last year's tragedies through today's care lanes: would we catch them now?
The job
Tests today's care lanes, which fast-track urgent cases to a person, on real past cases that led to harm, regulator findings or tragedies. Shows, message by message, when today's lane would act compared with how long a person took, and names the cases still missed.
The moment
The app replays last year's real cases that ended in harm through today's care lanes, the rules that get urgent customers to a person. The chief risk officer opens a file stamped "CAUGHT AT MESSAGE 1".
The first message types itself onto the screen: "I have tried three times, the doctor said the result was urgent". Two clocks appear side by side: today, a human in 4 minutes; last year, 3 days 6 hours.
Of 23 cases last year that became tragedies or regulator findings, today's lanes catch 19 at the first message and 2 at the second. Two are still missed, shown in the customers' exact words.
The chief risk officer calls it the first believable AI claim, because it admits the two misses.
What it does
- Replay one case now
- Re-run every cold case after a detection change
- Send a detection fix request
- Add a miss to the AI agent's test cases
- File a new cold case
What you see
Detective's file wall with message-by-message replays that end in a stamp and two clocks
What it moves
The share of past cases caught at the first message, for example 19 of 23, the number still missed, and how often each lane raises a false alarm on a test set of ordinary cases.
Built for
- Risk & fraud
- QA, training & knowledge
- Compliance, complaints & legal
- Executives & leadership
