Legal
AI Use
Draft version · last updated 14 September 2026
In short: AI helps structure and summarise. It suggests; people decide. AI suggestions are always labelled and never treated as confirmed on their own.
1.Plain-language summary
Qapacity may use artificial intelligence to help structure technical requirements, extract information from documents, suggest capability relationships, summarise information and assist with drafting.
AI-generated suggestions may be incomplete or incorrect. Where a decision depends on technical compatibility, safety, legal requirements or provider capability, users should rely on confirmed information and appropriate professional judgement.
2.Where AI is used today
When you describe a requirement in your own words, Qapacity suggests structured fields — such as objective, device, materials, operating conditions, deadline — and proposed criteria.
- Where AI is configured for the platform, these suggestions are produced using Anthropic’s Claude models through Anthropic’s API. The text of your requirement and Qapacity’s capability catalogue are sent to Anthropic for this purpose.
- Otherwise — or if the AI service is unavailable or declines a request — suggestions come from a rule-based extractor that runs within Qapacity and does not send your text to a third party.
Additional AI-assisted features may be introduced. This page will be updated when they are.
3.How AI suggestions are labelled
AI suggestions are clearly marked in the product with the label AI suggestion — review before accepting. A suggested criterion is not part of your requirement until a person reviews and accepts it. Nothing suggested by AI is silently saved as a confirmed requirement.
4.What AI does not do
AI on Qapacity must not silently:
- certify compatibility;
- invent facility capability;
- mark unknown information as confirmed;
- approve restricted access;
- override provider restrictions;
- approve legal or export-control compliance.
Facility matching is based on structured criteria and recorded facility information, not on AI similarity scores, and every assessment explains what matched, what did not and what is unknown.
5.Your data and AI providers
We send only the information needed to produce the suggestion. How Anthropic may process and retain API inputs is governed by its commercial terms. [Summary of the provider’s retention and training commitments to be confirmed on review.]
See the Privacy Notice for processors and international transfers.
6.Your responsibility
Review AI suggestions before accepting them. Do not include information in a requirement that you are not permitted to process with third-party services, such as information subject to security classification or export-control restrictions, unless you are permitted to do so.
Questions about this document? Use the contact page.