AI policy

1. Summary

• AI in Till is a copilot, not an autopilot. The agent drafts, analyses and recommends. You approve every action that touches the outside world.
• Every AI output is grounded in your own business data, your tracked competitors or clearly cited public sources. We do not invent figures.
• We use third-party AI providers (primarily Google Gemini) to power generation. Your data is sent to them solely to produce your results and is not used by them to train their models.
• AI-generated output is labelled in the product wherever it is shown, so you always know what came from the agent.
• You stay in control of corrections, deletions and the option to switch AI features off.

2. Our principles

1. Grounded over generic. The agent answers from your live business analysis, sales, team data, competitor audits and connected integrations. When it pulls from public sources (for example Google Places listings or competitor websites) it cites or links to them so you can verify.
2. Human approval for anything that leaves Till. The agent can draft a post, propose a competitor to track or suggest a price change, but it will not publish, send or charge anything without you clicking through.
3. Disclosed by default.AI-generated insights, drafts, audit summaries and recommendations are visibly labelled inside the product. Reports and content drafts carry an “AI generated” marker so you can pass that context on to anyone reading them.
4. Your data is yours. We send the minimum context needed to produce a result, and our agreements with AI providers prohibit them from using your prompts or outputs to train their underlying models.
5. Auditable. Significant AI actions (audits, content drafts, scheduled reports) are stored against your account so you can review, revisit or delete them at any time.

3. Where AI shows up in Till

a) The Till Agent (Ask Till)

The agent answers questions about your business by reading your live data through a set of tools: sales summaries, team rota, competitor audits, AI visibility scores, content drafts and your saved business analysis. Its answers are generated by Google Gemini using only the context relevant to your question. The agent does not have permission to mutate data on its own; tool calls that write or send anything require a confirmation step in the UI.

b) Business analysis

When you onboard or refresh your business analysis, Till pulls public information about your business (Google profile, website, reviews) and uses AI to summarise positioning, customer themes, content gaps and opportunities. The output is presented as analysis, not fact; we show source URLs so you can check the underlying material.

c) AI visibility audits

Visibility audits score how well an AI search assistant could understand and recommend your business. Scores are calculated from objective signals (profile completeness, NAP consistency, schema, content depth). AI is used to interpret content and write the human-readable summaries; the underlying signals are deterministic and re-runnable.

d) Content Studio

Content Studio generates draft copy (FAQs, About sections, services pages, posts) grounded in your business analysis. Drafts are stored against your account, never auto-published. You edit, approve and publish or copy them out into the surface of your choice.

e) Competitor intelligence

When you add a competitor, Till uses AI to summarise their positioning from publicly available data and to compare it to yours. Source listings and links are included so you can verify the underlying material before acting on the summary.

f) Recommendations and actions

Till proposes next steps (for example: a competitor to track, a content piece to write, a profile field to fill in). Each recommendation links back to the data that triggered it. Acting on a recommendation always requires a click from you.

4. What the AI does not do

The agent and the AI features inside Till do not, on their own:

• Publish or post content to your website, social channels, Google profile or any third-party platform.
• Send emails, SMS messages or push notifications to your customers or staff.
• Change pricing, charge customers, refund orders or move money through any payment provider.
• Edit, hire or fire staff records, contracts or shifts without an explicit confirmation step.
• Train any third-party model on your data, or share your data with AI providers for any purpose other than producing your result.
• Make legally or financially binding decisions in the meaning of UK GDPR Article 22 (solely automated decision-making with legal or similarly significant effects).

5. Data we send to AI providers

Till uses third-party generative AI to produce many of the outputs described above. The primary provider is Google Gemini via Google Cloud. We may add or substitute providers over time; we will update this page when we do.

When you use an AI feature, we send only the context relevant to your request, which can include:

• Your business profile (name, category, location, values).
• Aggregated sales, product and customer signals from your connected integrations.
• Summaries of your tracked competitors and their public listings.
• Visibility audit signals and the text content of pages we have audited.
• Your prompt or question to the agent and the recent conversation in that chat.

We do not send raw payment card numbers, staff passwords or any data we have not already collected under our privacy policy. Prompts and outputs are not used by the provider to train their models. For full detail on the lawful basis, retention and your rights over this data, see the privacy policy.

6. Accuracy and limitations

Generative AI can be confidently wrong. Till works hard to ground outputs in your own data and to cite public sources where it can, but we cannot guarantee that every analysis, summary or draft is free of mistakes. Treat AI outputs as a starting point for a human decision, not a final answer.

In particular:

• Numbers in the agent's replies are calculated from your live data at the moment of the question. If your integrations are out of date, the answer will be too.
• Visibility audits reflect what was visible to us at the time of the audit. Results can change as your site, profile or the wider web changes.
• Content drafts may include phrasing that needs adjustment for your tone or local context. Always read before you publish.
• Competitor summaries are built from public sources and may lag behind reality. We surface the underlying links so you can verify.

7. Your controls

• Review. Audits, content drafts and saved reports live in your account so you can re-read and re-run them.
• Edit and delete. You can edit drafts, delete reports and clear chats with the agent at any time from inside the product.
• Disconnect integrations. When you disconnect a payment, analytics or productivity integration, the agent stops sending that data to AI providers. See the privacy policy for retention timelines.
• Switch AI off. You can avoid AI features entirely by not using the agent, Content Studio or visibility audits. The core dashboards and integrations work without them.
• Delete your account. Deleting your account removes your data and the conversations the agent has had on your behalf, on the timeline set out in the privacy policy.

8. Disclosure when content leaves Till

Reports exported from Till and content drafted in Content Studio are marked as AI-assisted. If you publish AI-generated content from Till on your own website, social channels or marketing material, we encourage you to keep a similar disclosure visible to your readers, particularly for content covering health, legal or financial topics where AI disclosure is increasingly expected by regulators and search engines.

9. Reporting an AI issue

If the agent gives you an answer that is wrong, misleading or potentially harmful, please tell us. The fastest route is to email support@tillspot.com with a screenshot or a copy of the conversation. We review every report and will respond within two working days. Where appropriate we adjust the prompts, grounding sources or guardrails behind the feature and we will note material changes in the changelog below.

10. Changes to this policy

AI in Till is moving quickly and so is the regulation around it. We will update this page whenever we change the AI providers we use, add new AI features, expand what data is sent to providers or change the guardrails around what the agent can do without a human click. Material changes will be highlighted on the homepage of the product and, where they affect how your data is processed, in advance via email.

11. Contact

Questions, feedback or a request for more detail about how a specific AI feature works: support@tillspot.com.