Connect Google Vertex AI to Microsoft Agent 365

Use this guide to connect Google Vertex AI agents to Microsoft Agent 365.

Before you begin

  • Confirm that the Agent 365 AI administrator can create connections in the Microsoft 365 admin center.
  • Ask the platform administrator for the Google Cloud project ID and region.
  • Create a Google Cloud service account and a JSON key for that account.
  • Use the project and region where the agents are deployed.

Create the service account

Grant the service account the Vertex AI permissions required for agent discovery and management:

aiplatform.reasoningEngines.list
aiplatform.reasoningEngines.get
aiplatform.reasoningEngines.delete

The JSON key must contain the complete service-account key document, including the opening and closing braces. Don't enter only the private_key value.

Add observability permissions

To collect observability data, grant the same service account these roles:

Role Role ID Purpose
BigQuery Data Viewer roles/bigquery.dataViewer Read exported trace tables
BigQuery Job User roles/bigquery.jobUser Run queries to retrieve spans
Monitoring Viewer roles/monitoring.viewer Read request-count and latency metrics
Cloud Trace User roles/cloudtrace.user Read Cloud Trace spans

You need both BigQuery roles. Data Viewer permits reading the tables, while Job User permits running the queries that retrieve the results.

Configure Google Cloud telemetry

On the project where you deployed the agent:

  1. Enable the Vertex AI API.
  2. Enable the BigQuery API.
  3. Enable the Cloud Trace API.
  4. Enable the Cloud Monitoring API.
  5. Enable telemetry collection on the deployed agent.
  6. Enable Trace Analytics BigQuery export.
  7. Confirm that the trace_analytics_link dataset and its _AllSpans view exist.

Agent 365 can't enable telemetry collection or trace export for you.

Create the connection

  1. Open the Microsoft 365 admin center.
  2. Select Agents > All Agents.
  3. In Connected platforms, select Manage.
  4. Select + Connect a platform and choose Google Vertex AI.
  5. Enter the project ID, region, and complete service-account JSON key.
  6. Validate the credentials.
  7. Save the connection.
  8. Select Sync agents.

Verify the connection

Confirm that the expected agents are imported. A connection can validate and import agents while returning no telemetry if the project, region, roles, APIs, or trace export aren't configured correctly.

Troubleshoot

Symptom Likely cause Resolution
Agents import but no activity appears Observability isn't configured Complete the roles, API, telemetry, and trace-export steps.
No spans, but metrics appear BigQuery roles or trace export are missing Grant both BigQuery roles and enable trace export.
Queries fail although tables are readable roles/bigquery.jobUser is missing Grant BigQuery Job User to the service account.
No request-count or latency metrics Monitoring role or API is missing Grant Monitoring Viewer and enable Cloud Monitoring API.
_AllSpans doesn't exist Trace Analytics BigQuery export isn't enabled Enable the export and verify the dataset and view.
Agents import but telemetry is empty Connection scope doesn't match deployment scope Use the deployed agent's project and region.

For connection issues that aren't specific to Google Vertex AI, see Troubleshoot connected platforms.

  • Last updated on