Track · For developers & platform teams
Gemini Enterprise Agent Platform
Google Cloud's platform to build, scale, govern and optimize agents: Agent Runtime, Memory Bank, Agent Registry, Agent Gateway, policies and OpenTelemetry observability.
Module 1 Platform foundations
Module 2 Build
- 04 Build an ADK agent with Agents CLI From an empty folder to a project ready to deploy. Start →
- 05 Other frameworks on Agent Runtime LangChain, LangGraph, AG2, LlamaIndex, A2A and custom. Start →
- 06 Managed agents Agents Google runs for you, with a sandbox. Start →
- 07 Skill Registry Publishing and reusing agent skills. Start →
- 08 RAG Engine Corpora, ingestion, parsing, retrieval and reranking. Start →
- 09 Agent Retrieval (Vector Search 2.0) Collections, auto-embeddings, hybrid search. Start →
- 10 Agent Search Grounding agents on your enterprise data. Start →
Module 3 Deploy and scale with Agent Runtime
- 11 Deploy an ADK agent to Agent Runtime What gets packaged, built and created. Start →
- 12 Provision agents with Terraform Agents as infrastructure as code. Start →
- 13 Query a deployed agent Sessions, streaming and the SDK. Start →
- 14 Revisions and traffic Rolling out a new version safely. Start →
- 15 The runtime contract What Agent Runtime expects from your agent. Start →
- 16 Bidirectional streaming Live, two-way agent connections. Start →
- 17 Performance, scaling and networking Scaling settings and Private Service Connect. Start →
- 18 Autonomous agent scheduling Deferred, off-peak runs for long agent jobs. Start →
Module 4 Sessions, memory and feedback
- 19 Agent Platform Sessions Managed sessions for ADK and the API. Start →
- 20 Memory Bank: generating memories How conversations become long-term memories. Start →
- 21 Memory profiles, fetch and revisions Retrieving and versioning what's remembered. Start →
- 22 Feedback service Collecting user feedback on agent answers. Start →
- 23 Example Store Few-shot examples retrieved at run time. Start →
Module 5 Sandboxes
Module 6 Identity and access
Module 7 Agent Registry and topology
Module 8 Agent Gateway and policies
- 33 Agent Gateway overview One governed path for all agent traffic. Start →
- 34 Set up Agent Gateway Managed gateway, VPC connectivity and routing. Start →
- 35 Route Agent Runtime traffic through the gateway Putting deployed agents behind policy. Start →
- 36 Delegate authorization Letting an external system decide who gets in. Start →
- 37 Extensible policies Service extensions on the gateway. Start →
- 38 Unified Access Policies IAM policies and CEL conditions for agents. Start →
- 39 Semantic governance policies Rules about what agents may say and do. Start →
- 40 Token management policies Controlling token spend. Start →
- 41 Model Armor on the gateway Screening prompts and responses in transit. Start →
- 42 Dry runs and security findings Testing policies before enforcing them. Start →
- 43 Agent anomaly detection Spotting agents that behave unusually. Start →
Module 9 Observability with OpenTelemetry
- 44 OpenTelemetry for agents Traces, spans and attributes, from first principles. Start →
- 45 Set up tracing on Agent Runtime Sending agent traces to Cloud Trace. Start →
- 46 Logging and monitoring Logs, metrics and dashboards for deployed agents. Start →
- 47 Reading agent traces Finding the slow, failing or expensive step. Start →
- 48 Monitoring gateway traffic Gateway metrics and Model Armor spans. Start →
Module 10 Evaluate and optimize
- 49 Agent evaluation on Agent Platform Metrics, datasets and runs. Start →
- 50 Offline and simulated evaluation Testing before release. Start →
- 51 Online monitors and quality alerts Evaluating live traffic continuously. Start →
- 52 Failure clusters and prompt optimization From results to a better agent. Start →