1. Where AI is used in the Service
- Discovery and enrichment: classification, sector tagging, deadline extraction, and country detection on procurement notices.
- Qualification and scoring: automated fit scores against your organisation profile. Scores are indicators, not decisions.
- Drafting: generation of expressions of interest, methodology sections, cover letters, and proposal drafts, grounded in your knowledge base.
- Question and answer: chat interfaces that answer questions about a specific opportunity or your own uploaded content.
- CV and reference tailoring: reordering, summarising, and mapping of uploaded staff and project material against a specific opportunity.
2. Model providers and processing
Model calls are routed through vetted enterprise gateways. The current list of AI sub-processors is maintained on our Sub-processors page. All are engaged under written agreements that:
- Prohibit the use of your inputs and outputs to train shared or public models.
- Restrict retention of prompts and outputs to what is necessary to return the response and detect abuse (typically zero-retention or short abuse-monitoring windows).
- Require confidentiality and appropriate security.
Where possible we prefer UK or EU processing regions. Any transfer outside the UK is made on the safeguards described in the Privacy Policy and DPA.
3. What AI outputs are, and are not
LLM outputs are probabilistic. They may look confident even when they are wrong. They may omit important detail, invent citations or facts (hallucination), reflect training-data bias, be out of date, misinterpret ambiguous prompts, or produce plausible but legally-defective content.
Outputs are decision-support. They are not legal, financial, procurement, or professional advice. You must independently verify every output before relying on it, submitting it in a tender, sharing it with a client, publishing it, or communicating it to a donor or authority.
4. Your responsibilities
- Review and edit all AI-generated drafts before submission.
- Check that citations, statistics, dates, and names are correct against primary sources.
- Ensure the final submission complies with the procurement notice, the relevant integrity rules, and any framework agreement.
- Do not upload information you are not authorised to share, and do not upload special category or criminal-offence personal data without a lawful basis and safeguards.
- Consider whether disclosing your use of AI is required by the buyer's rules.
5. Human oversight
Every AI-generated draft or score is presented in the Service alongside controls that allow a human user in your Workspace to review, override, edit, or discard it. No opportunity is submitted to any authority solely on the basis of an AI output.
6. Training data and improvements
We do not use your Customer Content, prompts, or outputs to train shared or public AI models. We may use anonymised, aggregated telemetry to improve the Service (for example, rate of accepted suggestions) provided that no individual, Workspace, or Customer Content is identifiable.
7. Transparency and traceability
Where the Service produces a generated document, it records the model family used, the prompt version, the knowledge base sources retrieved, and the workspace user who accepted the output. This information is available on request.
8. Regulatory context
We monitor guidance from the UK Information Commissioner's Office, the AI Safety Institute, and other relevant bodies, and align our practices with the UK's principles- based approach to AI regulation. Where you or your buyer are subject to the EU AI Act, please tell us so we can support your obligations.
9. Reporting concerns
To report a harmful, biased, or otherwise concerning AI output, please write to laura@contracttenders.com. We investigate every report.
Questions about this document? Please write to laura@contracttenders.com. Contract Tenders is a service of Rhics Ltd, 150 City Road, London EC1V 2NX, United Kingdom.