Trust Centre

Runtime facts last reviewed 15 July 2026.

The five commitments

• Patient information is outside the product's intended use. AyrisIQ is for governance records and deliberately de-identified learning, not clinical records. The service blocks common identifiers and sensitive pilot AI paths, but no detector is a substitute for user redaction and contractual controls.
• Tenant and role rules are enforced at the database. Records and private files are scoped to the active organisation and membership. Independent production cross-tenant testing remains a launch gate before real customer data.
• AI suggests; your staff decide. Nothing AI-generated becomes part of your governance record until a named member of your team reviews and approves it. See AI safety.
• Primary storage and active AI are in London. Workspace data uses Supabase/AWS London; enabled general-governance AI uses Amazon Bedrock London. Global/EU supporting services and outstanding DPA evidence are listed on the data protection page.
• No behavioural advertising or session replay. We do use Sentry for scrubbed browser error reporting; the supplier and location are disclosed in the sub-processor register.

Find what you need

• Security — the controls protecting your workspace, verified against our codebase.
• Data protection — roles, data categories, locations, retention.
• AI safety — what AI does, what it may never do, and the human-review controls.
• Clinical safety — our DCB0129 position.
• Sub-processors — every supplier that touches service data.
• Compliance status — ICO, DSPT, Cyber Essentials, DTAC, testing — with dates, not badges we haven't earned.
• Evidence room — DPIA, DPA template and assurance packs, available to customers and prospects under NDA.

Questions, or something you need that isn't here? Contact us — governance questions get answered by the people who built the platform.