Answer users from your docs, know who they are when they ask, and let the agent act through your own API.
Train it on your docs, guides and release notes: PDF, Word, Markdown, HTML and more.
Pass the signed-in user’s ID and name to the widget, so every conversation is tied to an account.
Connect your API endpoints as actions, like checking a subscription. Sensitive ones can ask the user to confirm first.
Questions the agent could not answer are collected, so you can turn them into FAQs in a click.
Your help center and guides become the agent’s knowledge.
Point the agent at the endpoints it may call, with an audit log of every call.
Add the widget with the user’s identity, or build on the REST API.
Yes. Everything in the dashboard is available over a REST API with API-key authentication.
Only through actions you define. You can require confirmation, cap calls per session, and review every call in the audit log.
OpenAI and Anthropic models, with automatic fallback if one provider is down.