Skip to main content

​
What is a connector?

A connector is a secure, authenticated connection between Interact and an external data source. Without a connector, your agents have no data to work with — they’d be answering questions in a vacuum. Once you’ve set up a connector, you attach it to a subagent inside your agent. From that point on, the agent can query your data whenever you ask a question.

​
Supported data sources

Currently, Interact supports:

Google BigQuery

Connect your BigQuery project using a service account. Supports all BigQuery regions, standard SQL, and dataset-level access control.
Additional data sources (e.g. Snowflake, Redshift) are on the roadmap. Contact your Turntwo account manager if you have a specific requirement.

​
How connectors work

When an agent needs to answer a question, the Database Query Agent subagent:
  1. Receives the task from the orchestrator (e.g. “find total spend by campaign for last month”)
  2. Looks up your connector configuration to understand which tables are available
  3. Generates a SQL query scoped to the datasets you’ve whitelisted
  4. Sends the query to BigQuery using your service account credentials
  5. Returns the results to the orchestrator
This all happens in seconds, without any manual involvement.

​
Security

Interact takes several measures to keep your data and credentials safe:
Your service account JSON is encrypted using Google Cloud KMS before it’s stored in Interact’s database. Credentials are decrypted only at query execution time and are never exposed elsewhere.
When you create a connector, you choose exactly which datasets (and optionally which tables) agents are permitted to access. Agents cannot query anything outside that scope — even if they try.
You can set a maximum query cost per connector (in BigQuery billing units). Interact estimates query costs before running them and blocks any query that would exceed your limit, protecting you from unexpected BigQuery bills.
Connectors are always read-only. Interact only runs SELECT queries — it never writes to, modifies, or deletes data in your warehouse.

​
Managing multiple connectors

You can create multiple connectors — for example, one per client, one per region, or one per data environment (production vs. staging). Each connector has its own credential set, dataset whitelist, and cost limit. Agents can be configured to use any connector your organisation has set up. Set up your first connector →