Loup's AI features help managers write and find things faster. This page explains exactly what information they use, so you can be open with staff, the people you support and your regulator. If you'd rather not use AI at all, you can switch it off.
The principles
- People decide. AI only suggests. Nothing is saved, sent, changed or decided until someone on your team chooses to. It never makes decisions about care or employment.
- Only when asked. Nothing is sent to the AI unless someone presses an AI button or asks a question.
- As little personal information as possible. People you support are shown to the AI by first name and initial only.
- Same access rules. The AI can only use what the person using it is allowed to see. Inspectors with read-only access can't use AI.
- Not used to train AI. Loup uses Anthropic's Claude through its business service. Your information is used only to answer your request and is not used to train AI models.
- Your choice. A manager with full access can switch all AI features off in Settings → Which parts of the system do you use → AI features.
What each feature uses
| Feature | What is sent to the AI | Who can use it |
|---|---|---|
| Inspection-readiness review | Counts and percentages only — for example how many DBS checks are in date or how many incidents are open. No names and no written notes. | Managers with full access |
| Incident review | What your team has written in that incident (what happened, action taken), the person's first name and initial, and a short summary of their earlier incidents | Managers, and staff given AI access |
| Complaint letters | What is written in that complaint, the complainant's name (needed for the letter), and your agency and manager name | Managers, and staff given AI access |
| Monthly report summary | The figures in the report — no names of the people you support | Managers with full access |
| Ask your system | Only what each question needs, within that person's access — for example staff names with their training or check dates, or a short summary of incidents and complaints they can already see | Managers, and staff given AI access |
| Feedback action plans | Survey scores and written comments — never who said what | Managers, and staff given AI access |
"Staff given AI access" means office staff a manager has chosen to give AI access to in Users & Access, after reading a safety notice and recording why. The reason, who gave it and when are kept on that person's record.
Never sent to the AI
- DBS results and anything about criminal records
- Medication lists, medical notes and health details held on staff and client records
- Addresses, phone numbers, email addresses and next-of-kin details
- Full names of the people you support
- Uploaded documents, photos and certificates
- Passwords and login details
Records of AI use
Incident reviews and letters are kept with the record, showing who asked for them and when. Monthly reports note that the summary was drafted with AI and reviewed by the person who produced it. Questions asked in Ask your system are not stored.
What to tell people
Your agency decides how your records are used, so your own privacy notices should mention AI-assisted tools. You could use or adapt:
Details of the companies Loup uses to provide the service (including Anthropic), where information is processed and the safeguards in place are set out in Loup's data processing agreement and sub-processor list. For your own data protection impact assessment, ask your customer manager for Loup's DPIA template.
AI can make mistakes. Always check AI suggestions against what you know about your service and your own policies.
Still stuck? Contact support or speak to your customer manager.