Dokumentation

Data Connections

Connect external services and data sources for agent context, report data, and RAG integration.

Data Connections let you integrate external services into Fabric AI. Connected data sources provide context for agents, power report generation, and enable RAG-based document retrieval.

Overview

Data Connections support two authentication methods:

  • OAuth — For services like Google Drive, Notion, GitHub, and Slack
  • API Key / Credentials — For services like Snowflake and BigQuery

Each connection is scoped to your personal account or organization, with encrypted credential storage.

Setting Up a Connection

Go to Settings → Data Connections in the left sidebar.

Browse Available Providers

Switch to the Available Providers tab to see all supported services, organized by category.

Connect a Service

Click on a provider and follow the setup flow:

  • OAuth providers — Click "Connect" and complete the authorization flow in a popup window
  • API key providers — Enter your credentials directly

Verify the Connection

Once connected, the service appears in My Connections with a status indicator.

Connection Status

StatusDescription
ConnectedActive and ready to use
PendingSetup in progress
ExpiredToken needs refresh (OAuth connections)
ErrorConnection failed — check credentials

Multiple Connections

You can create multiple connections to the same provider. For example:

  • Connect two different Notion workspaces
  • Link multiple GitHub organizations
  • Access several Google Drive accounts

Each connection tracks its external workspace ID and name for easy identification.

Using Connections

With Agents

When creating agent instances, connect data sources in the Data Sources configuration:

  • Select which connections provide knowledge to the agent
  • OAuth connections use a simple "Connect" marker
  • The agent can then access data from these sources during execution

With Reports

Report templates specify required connections:

  • The template defines which integrations are needed
  • When creating a report instance, connect the required data sources
  • Reports pull data from these connections during execution

With RAG

Data connections can feed into the RAG pipeline:

  • Connected documents are processed and embedded
  • Agents and chat can then retrieve relevant context
  • Supports scheduled syncing for up-to-date data

Supported Providers

Providers are organized by category. Common categories include:

  • Productivity — Google Drive, Notion, Confluence
  • Development — GitHub, GitLab
  • Communication — Slack, Microsoft Teams
  • CRM — Salesforce, HubSpot
  • Analytics — Snowflake, BigQuery
  • Project Management — Linear, Jira

The exact list of available providers depends on your deployment configuration.

Security

  • Encrypted storage — All credentials (API keys, tokens) are encrypted at rest
  • Token refresh — OAuth tokens are automatically refreshed before expiration
  • Scoped access — Connections are isolated by user or organization
  • Revocable — Disconnect any service at any time

Next Steps