What we see, and what we never will.
The person who seems fine will only tell the truth in a check-in if they're sure nobody's watching. So we built MANA so that nobody is.
Who sees what
| What | Who can see it |
|---|---|
| Your answer to the weekly check-in | Nobody. Not your manager, not HR, not the volunteers, not us |
| That you asked for someone to check in | Only the volunteer you picked, as a nudge to say hello. Nobody else, and it's never reported |
| What you say in a conversation with a volunteer | It stays between you, unless you're at risk and they follow your workplace's written escalation route |
| Team results | The responsible person, as anonymous numbers, and only when five or more people answered |
| Whether volunteers are asking | The responsible person, as a percentage across all volunteers |
| Your course badge details | You and us. Your employer only if you say so |
How someone's week is going counts as health data, so we treat it that way: a full data protection impact assessment, UK or EU data agreements with every supplier, and nothing kept longer than it needs to be.
How we use automation and AI
We automate the admin so our people can spend their time on people. The rules we hold ourselves to:
- AI never talks to anyone about how they're doing. Every check-in message is fixed wording written by a person.
- Check-in answers never go into an AI model. If AI touches check-in data at all, it's anonymous counts from groups of five or more.
- AI drafts and a person decides, on anything that's a judgement, a promise, or about a workplace's standing.
- AI never assesses risk, never awards the mark, and never replies to anything that sounds like a crisis.
- Every tool we use has a UK or EU data agreement and doesn't train on our data. If it can't promise that, it doesn't touch customer data.
- We publish the list of tools we use.