Trust center

The boundary is visible.

An assistant people can question needs boundaries people can read. Here they are.

Knowledge

Sources are inspectable

Everything the assistant can say traces to sources you can read. There is no hidden index and no mystery context.

Answers

Citation, always

Every answer names its sources. An answer that cannot point at its material does not ship as an answer.

Corrections

Humans override

A correction from a human becomes part of the knowledge. The assistant never overrules the person maintaining it.

Visitors

No automatic email

Names and emails left in the public chat stay in your browser. DialogueBase sends nothing; follow-up is yours, by export, on your own email.

Tracking

No trackers

No analytics SDKs, no session recording, no third-party pixels.

Deletion

Local and total

Sources, conversations, test cases, and contacts live in local storage. Deleting the workspace removes them completely.

Audit

Quality you can rerun

Test cases turn quality into a check, not a feeling. Edit a source, rerun the suite, and see exactly what changed - before a visitor finds it.