Trust and data protection

The kinds of data involved

Blengi handles your own content, the conversations your visitors have, and the details a visitor chooses to give you. It helps to keep them apart. Your content is the pages and sitemap of your website, plus any PDFs, Word files, Google Docs, Notion pages, pasted text and SQL databases you connect as knowledge sources. Blengi indexes that material so the assistant can answer from it. Sources are re-crawled automatically, so the answers follow your content, and a source can be re-indexed on demand when a change needs to land.

Conversation data is the questions visitors ask and the answers they are given. The rest is whatever a visitor volunteers, such as a name, an email address or a phone number offered when the assistant captures a lead or books a call through Calendly.

There is no other category. The assistant answers only from the content you connect, and says when it does not know, so it has no reason to reach past your material into the open internet. It repeats what your indexed pages say. Anything about a specific order, or about what is available right now, goes to a person.

What that means in practice

• The knowledge base holds only the sources you added, and you can remove or replace any of them.
• Answers come from that material, and the assistant says when it does not know rather than filling the gap.
• Questions it could not answer surface as content gaps, so a person on your side fixes the page behind them once.
• Per-page and per-path rules decide where the assistant appears, so it stays off pages where it does not belong.
• Agencies get a separate knowledge base per client under their own brand, with the same isolation between each one.