Version: 1.0 Effective date: 9 October 2026
This policy explains how KF Premium LTD ("XEWI", "we") uses artificial intelligence in the XEWI platform: what it does, what it does not do, how your data is handled, and what you are responsible for when you use AI-generated output. It forms part of our Terms of Service.
1. Our principles
- You stay in control. AI prepares; you decide. No AI-prepared change that spends your money or changes your listings reaches Amazon without your approval.
- Hypotheses, not facts. AI-assisted analysis is presented as estimates and hypotheses, with its sources and confidence where we have them. It is never presented as fact.
- Honest gaps. When data is missing or too thin, we say so rather than let a model fill the gap.
- Your data is used to serve you. We send AI providers only what a task needs, under terms that do not allow them to train their models on your data.
- No scraping of review text. We do not scrape Amazon review text. Customer voice analysis uses Amazon's own Customer Feedback data, obtained through Amazon's official API.
2. Where XEWI uses AI
| Feature | What AI does | Human control |
|---|---|---|
| Listing copy | Drafts titles, bullet points and descriptions using your product data and market context | Shown next to your current listing as a before/after comparison; applied only after you approve, and can be rolled back |
| Product and lifestyle images | Generates candidate images | Applied to a listing only after you approve |
| Opportunity briefs, strike briefs and verdict explanations | Summarises statistical findings in plain language | Informational. Acting on them goes through the approval queue |
| Assistant (chat) | Answers questions about your data and XEWI's analysis, can search the web, and can suggest actions | Suggested actions go to the approval queue |
| Competitor briefs | Summarises where a competitor looks vulnerable, from public data | Informational hypotheses; acting on them goes through the approval queue |
| Classification and similarity | Groups search terms and feedback themes, labels products, and finds similar past situations | Informational; affects what we show you |
Most of XEWI's numbers are not produced by generative AI. Sales ranges, revenue floors, margins, fees and funnel filters come from statistical models and from data supplied by Amazon and our data providers. Generative AI is used to explain and draft, not to invent figures. Where a figure is estimated, it is shown as a range with its basis.
3. Human oversight
3.1 AI-prepared money and brand actions always wait for approval. Bid changes, negative keywords, listing and image changes, and workflow actions that XEWI's AI prepares are placed in an approval queue. They are submitted to Amazon only when you, or a team member you authorise, approve them.
3.1a Separately from AI, some rule-based automations (for example, a repricing rule you set to apply without approval) act on Amazon only when you expressly switch them on. These are covered by clause 5.3 of the Terms of Service.
3.2 Where approving only takes you to Amazon to make the change yourself (rather than XEWI making it), the Service tells you so.
3.3 You can reject or roll back AI-generated listing changes, and we record what was proposed, approved and applied, so that you have an audit trail.
4. Your data and AI providers
4.1 What we send. We send AI providers only the information needed for the task. For example:
- product attributes and listing content (yours and public competitor listings);
- market statistics and Amazon's aggregated Customer Feedback themes;
- the messages you type into the assistant, and recent conversation history; and
- your instructions.
We do not send your customers' personal data (such as buyer names, email addresses, postal addresses or VAT numbers) to AI providers.
4.1a Web search. When the assistant needs current information from the web, your question is sent to our web search provider as a search query. Avoid including personal or confidential information in questions you ask the assistant.
4.2 No training on your data. We use AI providers under commercial or enterprise terms that do not permit them to use your inputs or outputs to train their models.
4.3 Where it is processed. Some providers process data outside the UK. Where they do, we rely on the transfer safeguards described in our Privacy Notice.
4.4 Our own use. We do not use your data to train general-purpose AI models. We may use de-identified, aggregated signals, such as whether proposed changes were approved and how they performed, to improve XEWI's recommendations for you. We only share them across customers in aggregated form that identifies no one, as described in the Terms of Service.
4.5 The current list of AI providers is in the Sub-processor List.
5. Limits of AI output
AI output can be wrong, incomplete, outdated or inconsistent. In particular:
- Listing copy can contain claims that are inaccurate, unsupported, or not allowed by Amazon's policies or advertising and consumer-protection law (for example, health, safety, "best" or comparative claims);
- Images can misrepresent a product's size, colour, contents or use, or resemble third-party brands, trademarks or people;
- Explanations can overstate confidence or misread the underlying data; and
- Statements about competitors are hypotheses drawn from public data, not facts about those businesses.
6. Your responsibilities
When you use AI-generated output, you agree to:
- Review before you approve or publish. Check that listing copy and images are accurate for your product, comply with Amazon's policies and applicable law, and do not infringe anyone's intellectual property.
- Not present hypotheses as facts. Do not publish AI analysis about identifiable competitors as statements of fact.
- Label AI content where required. Disclose AI-generated or AI-altered content where the law or a platform's rules require it.
- Not misuse it. Do not use XEWI's AI features to generate misleading product claims, fake reviews or endorsements, content that impersonates people or brands, or anything prohibited by our Acceptable Use Policy.
- Keep your own judgement. Treat AI output as a draft or input to your decision, not as professional advice.
7. Regulation
XEWI's AI features are decision-support and content-drafting tools for businesses. We design them to meet transparency expectations: we label AI-generated drafts in the Service, and we keep a person in control of consequential actions. We keep this policy under review as UK and EU AI rules develop.
8. Cost limits
To keep AI use sustainable, we apply per-account limits on AI usage. When a limit is reached, non-essential AI tasks may be delayed until the next period. Essential functions and your approval queue are not affected.
9. Changes and contact
We may update this policy as our features and providers change, and we will notify you of material changes in line with the Terms of Service. Questions: hello@xewi.app. Privacy questions: privacy@xewi.app.