AI Agents admin
Create and manage the retrieval copilots that answer your users — grounded in your own documents, spreadsheets, agency directory, web pages and FAQs.
- Five knowledge types per agent
- Background ingestion with live progress
- Streaming, source-grounded answers
AI Agents
Create and manage the AI agents that answer your users from your own knowledge base.
| Agent | Knowledge | Identifier | Status | Updated | Actions |
|---|
No agents match these filters.
No agents yet
Create your first agent, then feed it documents, spreadsheets, agency data, URLs and FAQs.
Create a new agent
Pick a starting persona, then name your agent. Everything here can be changed later.
Agent
No dataIdentifier
Identity
Name, description and the persona that shapes every answer.
Documents
PDF, TXT, DOCX or MD files. Each is chunked and embedded for retrieval.
Tabular data
CSV or XLSX spreadsheets — one row becomes one retrievable record.
Agency data
CSV or XLSX where each row is an agency.
Do not upload ordinary spreadsheets here — use Tabular data for those. Agency ingestion summarises each field group with the LLM, so it is considerably slower than the other pipelines.
Web pages
Public URLs are fetched, chunked and embedded like documents.
FAQs
Curated pairs. A confident FAQ match is returned verbatim, ahead of retrieval.
Deployments
Embed this agent on a website. Each site gets its own token, which you can pause, rotate or revoke on its own.
The token is not a secret — it ships in the page source of a public website, where anyone can read it. What binds it is the domain: a browser on any other site is refused. Treat everything this agent knows as readable by anyone who visits the site you deploy it to.
Chat
Talk to the live agent exactly as an end user would.
User sessions
Every conversation your agents have held — including ones their owners deleted, which are archived here until you remove them forever.
| Session | Agent | Owner | Messages | Status | Last activity | Actions |
|---|
No sessions match these filters.