Skip to main content

​
Prerequisites

Before you begin, make sure you have:
  • Access to a Turntwo Interact workspace (if you don’t have one, contact your Turntwo account manager)
  • A Google BigQuery project with at least one dataset you want to analyse
  • A BigQuery service account with the BigQuery Data Viewer and BigQuery Job User roles
If your organisation already has a connector set up, you can skip Step 1 and go straight to building your agent.

​
Step 1: Connect your BigQuery data

1

Open Connectors in Settings

In the left sidebar, go to Settings → Connectors and click Add connector.
2

Choose Google BigQuery

Select Google BigQuery from the list of available data sources.
3

Upload your service account credentials

Paste in your service account JSON credentials and enter your Google Cloud Project ID.
Your credentials are encrypted at rest using Google Cloud KMS — they are never stored in plain text.
4

Select datasets to expose

Choose which BigQuery datasets your agents are allowed to query. You can whitelist specific tables if needed.
5

Save and test the connection

Click Save connector. Interact will verify the connection and list the accessible tables.

Full BigQuery setup guide

Detailed instructions including IAM permissions, dataset whitelisting, and query cost limits.

​
Step 2: Build your first agent

1

Go to Agents

From the sidebar, click Agents and then Create agent.
2

Give your agent a name and description

Use something descriptive like “Campaign Performance Analyst”. The description helps the agent understand its purpose.
3

Add your connector

In the Subagents section, add a Database Query Agent and attach the BigQuery connector you just created.
4

Add context (optional)

In the agent’s Global context field, describe your data model in plain language — for example:
“The events table tracks ad impressions and clicks. The campaigns table contains campaign metadata including budget and flight dates.”
This helps the agent write accurate queries without needing to be corrected. If your organisation has already filled in shared organization context, enable that checkbox too instead of repeating the same facts here.
5

Save your agent

Click Save. Your agent is ready to use.

Learn more about agents

Full guide to configuring orchestrators, subagents, model selection, and tools.

​
Step 3: Run your first analysis

1

Open the agent chat

Click on your newly created agent to open the chat interface.
2

Ask a question in plain language

Type a question about your data. For example:
“How many impressions did each campaign get last week?”
The agent will translate your question into a SQL query, run it against BigQuery, and return the results.
3

Explore the results

Results appear as a table, chart or report directly in the chat. You can ask follow-up questions, or say how you want the answer presented:
“Can you break that down by device type and show it as a bar chart?”
See which visualisation fits which question →

​
What’s next?

Schedule a job

Automate your analysis to run on a daily, weekly, or custom schedule.

Set up notifications

Get Slack or email alerts when your agent finishes a run.

Build a report

Create structured, repeatable reports from your agent’s output.

Use the API

Trigger agent runs programmatically from your own systems.