AI, data & privacy
HappierBiz has two AI features, the Help assistant (the ? panel, available on every plan) and the full AI assistant (a Business-plan feature that can answer from your own data). This page explains, honestly, what data they use, what is stored, and what the AI provider does with it, so you and your team can answer the question "is my data safe with the AI?" truthfully.
In one line: the Help assistant never receives your business data; the full assistant only ever receives a minimal, de-identified summary; HappierBiz does not store your conversations; and questions are sent to Anthropic (the maker of Claude), which does not train its models on them.
What each AI feature sends
| Feature | What is sent to the AI |
|---|---|
| Help assistant (all plans) | Product knowledge about the app (how-to / where-is), your business name, and what you type into the chat. No bookings, customers, money or team data. |
| Full AI assistant (Business) | The above, plus a de-identified snapshot across the tools you've switched on (schedule, money owing, customers, team, leave, certifications, safety, assets, stock, morale): totals, counts and first names only, never emails, phone numbers, surnames, addresses, health/incident detail or document contents. Plus what you type. |
| Full AI assistant, note lookup (Business) | When a question needs a specific detail the summary doesn't hold, the assistant can search your own internal notes, handover notes, notes written on a customer's profile, past team announcements, form submissions, safety/maintenance notes, and the reason on a leave request, and only the matching excerpts are sent to answer it. |
The full assistant's snapshot is built to be deliberately minimal, it's aggregates and first names so it can be useful ("you have 3 bookings today, next is Sarah at 2pm") without exposing contact details.
Looking up your own notes
The full assistant can answer deeper questions ("what did yesterday's handover say?", "what's noted on Mia's profile?", "did we post anything to the team about Saturday?") by searching your business's own written records. A few important limits keep this safe:
- It only searches when a question needs it, it isn't constantly reading your notes, and it never bulk-uploads them.
- You choose what it may read. Under Settings → Modules → "What the AI assistant may read" you can switch each source on or off for the assistant independently. A source switched off there is never searched, even though the underlying module stays on for your own use.
- It also only searches sources whose module is switched on (handover notes and announcements need Team comms; customer notes need Customers; form submissions need Forms; safety/maintenance notes need Safety; leave reasons need Leave). Turn a module off and that source is invisible to it.
- Only the handful of matching excerpts are sent, not the whole table.
- It only ever reads your own business's records (tenant-isolated), never another business's.
This is the assistant reading back your own notes to you, on request, the same records you can already open in the app.
What HappierBiz stores
- Not your conversations. HappierBiz keeps only a count of how many AI questions you've used each month (to apply your allowance) and the token totals. The questions and answers themselves are not saved in HappierBiz, they live only in your browser while the panel is open and are gone when you refresh.
- No transcripts, no chat history database.
What the AI provider (Anthropic) does
Your questions are sent over a secure, encrypted connection to Anthropic, the company that makes Claude (the AI model HappierBiz uses). Per Anthropic's commercial terms:
- Your inputs and outputs are not used to train Anthropic's models.
- They are retained only briefly, for safety and abuse monitoring, and then deleted (content flagged for a policy violation may be kept longer).
- Data is encrypted in transit and at rest.
The authoritative source for these terms is your business's privacy policy and Anthropic's Commercial Terms of Service / Trust Center. This page is a plain-English summary, not the contract.
The one thing to be careful about
The AI can't stop you typing something into the chat. Whatever you type is sent to the AI provider. So:
- Don't paste customers' personal details (full names, emails, phone numbers, card or health information) into the chat. You don't need to, the assistant already has the safe summary it needs.
- Treat the chat like sending an email to a trusted supplier: fine for "how do I set up deposits?", not the place for a customer's private record.
Frequently asked
Does the AI see my business data? The Help assistant: no, only your business name and your question. The full assistant (Business): only a minimal summary (totals + first names), never contact details, plus, when you ask something specific, the matching excerpts from your own internal notes (see Looking up your own notes above).
Does Claude/Anthropic keep my questions or train on them? Not for training. Anthropic doesn't use commercial-API data to train its models, and retains it only briefly for safety before deleting it.
Does HappierBiz store my conversations? No, only a count of questions used. The chat isn't saved.
Can I turn the AI off? The full assistant is a module you switch on/off in Settings → Modules. The Help assistant is part of the product; if you'd prefer it not be used, ask your team not to type sensitive details into it.
Related: AI assistant · The dashboard & navigation · FAQs