HomeTerms

Terms

Analytics and AI Development Terms

Draft for legal review. Not in force.

DRAFT, NEEDS LEGAL REVIEW. Not yet in force. Not a live offer.

For analytics, reporting, dashboards, AI development, model building and AI-enabled services. Separate from Data terms.

1. What we do

  1. Analytics: reports, dashboards, forecasts, scoring and insight built from data you give us or data we licence.
  2. AI development: design, building, tuning and integration of models, assistants and automation for your use, under a statement of work (SOW).

2. Inputs and your duties

  1. You are responsible for the quality, lawfulness and rights of data you provide. Poor inputs give poor outputs, and we are not liable for results that follow.
  2. You give us the access, domain knowledge and sign-offs described in the SOW, on time.
  3. You must not ask us to build systems for unlawful surveillance, discrimination, deception, weapons, or any use that breaks the law or human rights.

3. Outputs and their limits

  1. Outputs are statistical or model-based. They can be wrong, incomplete or out of date. They support human decisions; they do not replace them.
  2. You are responsible for checking outputs before relying on them, particularly in finance, safety, hiring, credit, health or legal matters. Use in high-risk decisions needs a written risk review and a human in the loop.
  3. We state measured accuracy, test method and known limits in the delivery note. We do not guarantee accuracy in live use, because live data differs from test data.
  4. Third-party models and tools: where the work uses third-party models or services, their terms and limits apply, and we tell you at quotation.

4. Ownership and use of models

  1. Custom models and code built for you with your data: when paid in full, you own the trained weights and code specific to you. Our base models, tools, pipelines and know-how stay ours, and you get a perpetual internal-use licence to them as embedded in the deliverable.
  2. Your data is not used to train models for anyone else. We may keep general learning that does not contain or reveal your data.
  3. Reports and dashboards are licensed for your internal use. Publishing or selling them needs our written consent.
  4. Data handling in analytics follows the Data terms, including the Data Processing Terms.

5. Transparency, fairness and safety

  1. Where the system affects people, we help you set up notices, human review, and a way for people to challenge a decision, as the law and good practice require.
  2. We test for obvious bias and security faults before delivery and describe the tests. Continuing monitoring after delivery is a separate service.
  3. Generated content (text, images, code) may mirror training material and may need your own review for rights and accuracy before use.

6. Delivery, acceptance and warranty

  1. Work runs in stages with acceptance tests stated in the SOW. Acceptance and warranty follow the Software terms (10 working days to test; 90 days to fix reproducible defects against the SOW).
  2. Model drift, new data patterns and changes by third-party services are not defects. Retraining and monitoring are priced separately.

7. Pricing models

  1. Fixed price per stage, time and materials at the rate card, or subscription for managed analytics. Third-party compute, data and tool costs are listed and passed on at cost, unless the quote includes them.

8. Liability

  1. General Terms limits apply. We are not liable for decisions made or losses arising from relying on outputs, other than as the SOW says.