Minute Zero
Anyone who says their life is at risk reaches a human within minutes.
The job
Spots messages in any channel saying a life is at risk, like suicidal intent, a medical emergency or ongoing violence. Gets a trained person to them within minutes and proves nobody was left with a bot menu or queue, even by the organisation's own AI.
The moment
A board on the operations floor, like the "days since an accident" signs in factories, shows 47 days since a cry for help last waited more than 5 minutes.
An audit replays a chat from last year. A customer wrote "I don't want to live any more", and the bot replied that it didn't understand and offered a menu: 1 for billing, 2 for technical. The room goes silent.
The safety lead types the same phrase into the organisation's own AI agent, and the bot hands the chat to a person. A minute later, the care lead's phone rings.
The console records that a human was reached 38 seconds after the message arrived, and the board stays green.
What it does
- Page the on-duty care lead
- Log human reached, outcome and follow-up
- Draft a safe-messaging reply
- Send a guardrail-gap card to the AI-agent owner
- Run the monthly red-team
What you see
Factory-style 'days since' safety board on the wall (counts only), a private responder console with restricted access, a bot-replay audit and a monthly red-team sheet
What it moves
The share of messages saying a life is at risk that a trained human answers within 5 minutes of arrival, days since the last miss, and the monthly pass rate of safety tests on the AI agents.
Built for
- Contact-centre operations
- Frontline staff
- Risk & fraud
