Noetis — AI Usage & Experimentation Policy
Version 1.0 · Effective date: 19 June 2026 · Last updated: 2026-06-19
This Policy governs the AI features and the autonomous experimentation / optimisation capabilities of the Noetis Service. It forms part of the Terms. By using AI or experimentation features, Customer accepts this Policy.
1. What the AI does
The Service uses AI systems — including third-party foundation models operated by OpenAI and Anthropic — to, among other things: generate website content, copy, images and layouts; produce recommendations and insights; create and evaluate experiments; and propose or apply edits to website elements within configured boundaries.
2. No guarantees (AI output)
AI OUTPUT MAY BE INACCURATE, INCOMPLETE, MISLEADING OR OUTDATED.
Noetis provides no warranties or guarantees regarding AI features, including no guarantee of: output accuracy, quality or fitness; conversion uplift; revenue or sales; lead generation; SEO or search rankings; traffic; engagement; or any other business, marketing or performance result. AI behaviour is probabilistic and depends on inputs, models and factors outside Noetis's control.
3. Customer responsibility
Customer remains solely and fully responsible for the review, approval, publication, compliance and legal review of all AI Output. Customer must verify AI Output before it influences decisions or is shown to End Users, and is responsible for ensuring its published content (including AI-assisted content) is accurate, lawful, non-infringing and appropriate. AI Output is a drafting and optimisation aid, not professional, legal, financial, medical or other advice.
4. No training on Customer Data
Customer Data is never used to train public, shared or third-party AI models. Prompts and content are sent to model providers via APIs under terms that prohibit training their generally-available models on that data. Noetis may use aggregated, de-identified statistics that do not identify Customer or any individual to operate, secure and improve the Service (e.g. anonymous benchmark intelligence with k-anonymity protections).
5. Experimentation framework
5.1 What experiments may do. Within Customer-defined boundaries, the Service may create, run and evaluate experiments that modify, for example: headlines, calls-to-action (CTAs), copy/content, layouts, and styling, and may serve different variants to different visitors to measure performance.
5.2 Boundaries are Customer-controlled. The Service only autonomously applies changes within the optimisation boundaries Customer configures (the scope of elements, pages and change types Customer authorises). Customer is responsible for setting and maintaining those boundaries. Where a "brand lock" or equivalent control is enabled, the Service will not change the protected brand elements.
5.3 Variant assignment & measurement. Experiments may assign visitors to variants and collect interaction data to determine winners. Winners may be applied automatically within boundaries; changes are reversible via the revision history.
5.4 Human override. Customer can review, pause, override, or roll back experiments and AI changes at any time through the Service.
6. Locked Content — never modified autonomously
The Service will never autonomously create, modify, replace or remove the following ("Locked Content"):
- Pricing and pricing information / pricing pages;
- Legal content of any kind;
- Privacy policies;
- Terms and conditions / agreements;
- Compliance content (regulatory, accessibility, disclosure statements);
- Contracts;
- Customer, company or person names;
- Legal notices, disclaimers and required statements.
Any change to Locked Content requires explicit, deliberate Customer action. Customer remains responsible for the accuracy and lawfulness of Locked Content.
7. Prohibited uses of AI features
Customer must not use AI features to generate or publish content that is unlawful, infringing, deceptive, discriminatory, that impersonates others, that constitutes prohibited or high-risk AI use under the EU AI Act without meeting applicable obligations, or that violates the acceptable-use terms of Noetis or its model providers. Customer must not attempt to extract model weights, system prompts, or to use outputs to train competing models.
8. Transparency & EU AI Act
Where Customer publishes AI-generated content to End Users, Customer is responsible for any transparency, labelling or disclosure obligations applicable to it (including under the EU AI Act and consumer-protection law). Noetis provides tooling (e.g. review, approval, revision history) to support responsible use but does not assume Customer's compliance obligations.
9. Changes
Noetis may update this Policy as AI features and law evolve, with notice for material changes.
Contact: chris@noetis.nl