Start in AI Studio.

Query connected data, draft model tasks and audience rules, set up attribution, and review results in dashboards. Publishing and write actions follow the permissions and review steps of the workflow.

What you can do

It is you, never more

The assistant acts as the signed-in person. A database, workflow or dashboard you cannot read does not exist for it, and its name never appears. Access is checked again on every message and every tool call, so losing access takes effect at once.

It reads rows, under your column rules

Queries are read-only and run through the same access, column tags and masks the console applies. Rows stay inside the chat that read them: logs, audit records and metrics keep references, never values.

Dashboards, in the console

Ask for a view of what a run produced and you get a dashboard in the console, with properly formatted figures and named charts. Chats and the dashboards they produce are private to whoever made them, and theirs to delete outright. A dashboard can offer a follow-up question; clicking it fills your message box, and you decide whether to send it.

It sets workflows up with you

The assistant can draft a schema, a use case, an attribution setup or an audience from your data and hand it over for an admin to apply. It never publishes, never starts a run that costs money, and never writes to your systems on its own.

Your own connectors

Connect Snowflake, Databricks, HubSpot, Salesforce, BigQuery, GitHub, Linear or any https MCP server from Settings, with your own access token or by signing in. Each call waits for your approval, admins can limit it to listed hosts, and you can switch a connector off for a single chat.

Admins hold the switches

Whether the model may read rows is on by default and an admin can switch it off, which leaves the assistant able to discuss the shape of your data but not its values. Which model provider is used — including your own key — what it may spend, and how long chats are kept are admin settings too. You delete your own chats and dashboards outright.

Setup and requirements

Steps

  1. AI Studio is open to members by default; an admin can narrow it to admins.
  2. Ask a question about a workflow or a database in plain words.
  3. Read the SQL it wrote and the rows it got back, and ask for a dashboard.
  4. Connect your own tools from Settings › Connectors, approving each call.

What you need

  • A workflow that has produced rows, or a database you can read
  • An admin decision on which model provider is used, and whether to leave row reading on

The assistant reads row values by default, and those rows are sent to the model provider your organisation uses. The switch is an admin's and turning it off leaves the assistant able to discuss the shape of your data but not its values. Ask us which providers are available to you, and how they are covered, before your team starts using it.

Build an Agentic Worker for your team.

Tell us which workflow you want to hand off.

Book a demo