Secure Agents Built with Google ADK and Vertex AI
Trust3 AI integrates with Google ADK and Vertex AI Agent Builder to apply authorization across agent workflows. Trace connected tool use, secure delegation and protect access to enterprise data.
Authorize the actions an agent chooses
An agent can select a tool, share context and trigger a business action in the same workflow. Security needs to follow that sequence so a valid credential does not become permission to perform any action the agent can compose.
Agent Discovery Observability and Security
- Discover Bring agents connected through supported integrations and instrumentation into the Trust3 AI inventory. Record ownership, operating identity, approved scope and reachable tools and data.
- Observe Capture supported prompt, tool and delegation telemetry with policy context. Investigate where data entered the workflow and where the agent sent or used its output.
- Secure Evaluate supported requests against centrally managed policy. Constrain tool access and A2A handoffs, with identity-aware authorization and runtime guardrails for sensitive workflows.
Extend security to BigQuery and connected sources
Trust3 AI combines agent controls with supported native data authorization on Google Cloud. BigQuery and Cloud Storage integrations form part of the broader data-access layer, alongside connected platforms such as Snowflake and Databricks. Available controls depend on the source and access path.
Illustrative workflow
An ADK agent retrieves approved customer context, then calls a downstream tool. The demonstration checks whether the user, task and tool are authorized and records the policy decision alongside the action.
See an ADK workflow with tool authorization, delegation traces and protected data access.
