Built on ethical AI principles
We build tools to support professional work, not replace professional judgement. A person checks every AI output before use. We support compliance; we are not endorsed by a regulator.
Fair and ethical design
We design AI-assisted workflows with fairness and professional obligations in mind. Outputs require appropriate human scrutiny, especially where they could affect people.
Safe and robust AI
We test workflows for reliability and surface uncertainty for review. The relevant checks and thresholds depend on the product and task.
Responsible AI
Automation supports professional work; it does not replace professional judgement. People review and approve consequential outputs in their own workflow.
Explainable AI
We aim to make outputs reviewable, with sources or traceability where the workflow allows, so errors can be identified and corrected.
Trustworthy AI
We minimise data and apply safeguards appropriate to each workflow. Some products use external processors; the product's privacy and security information explains the details and exceptions.
Professional judgement stays human. We take account of relevant ICO guidance and UK regulatory expectations when designing AI-assisted tools. Refer to each product's policy for its specific safeguards.