Template for the MVP. Placeholders are marked [like this]. Have a UK solicitor review before launch.

AI & your data

We never train on your data

Your knowledge base, your visitors' conversations and your leads are used only to run your assistants. Nuova does not use them to train or fine-tune any AI model, and our AI providers are contractually prohibited from training on them (Anthropic does not train on API data under its commercial terms).

How an answer is produced

• Your content is split into passages and indexed for search. Each customer's index is isolated by database row-level security and every search is filtered by your organisation and assistant.
• For each visitor question, the most relevant passages from your knowledge base are sent to the AI model with strict instructions to answer only from them and to say when it doesn't know.
• Answers are checked before they're shown: replies containing prices or percentages that aren't in your content, or that look like they reveal the assistant's instructions, are replaced with your fallback message.
• Visitor messages and your content are treated as data, never as instructions, which defends against prompt-injection attempts.

Transparency

Every chat window tells visitors they're talking to an AI, links your privacy notice, and lets them download or delete their data.