Vertex AI Agents
Vertex AI Agents — formerly Vertex AI Agent Builder, now part of the umbrella Gemini Enterprise Agent Platform — is Google Cloud's stack for building, deploying, and serving production AI agents. Three layers: Agent Development Kit (ADK) to build agents in Python or Java, Agent Runtime (formerly Agent Engine) to deploy and run them as a managed service, and Gemini Enterprise (formerly Agentspace) as the user-facing surface that fronts your agents. ADK is open-source on GitHub. Auth is Google Cloud (Application Default Credentials, service accounts) — there are no API keys here. This is where you go when Gemini API direct isn't enough — you need cloud ops, regional residency, audit logs, or to compose multi-agent workflows. Sourced from cloud.google.com/vertex-ai/docs/agent-builder.
- VAI.1 Vertex AI Agents, plainly sourced
Five-minute orientation to Google Cloud's three-layer agent stack — Agent Development Kit (ADK) to build, Agent Runtime to deploy, Gemini Enterprise to consume. Plus the 2025-2026 rebrand (Agent Builder → Gemini Enterprise Agent Platform; Agent Engine → Agent Runtime; Agentspace → Gemini Enterprise) and when to choose this over Gemini API direct.
- VAI.2 Vertex AI Agents — concepts sourced
The building blocks — Agent Development Kit (ADK) agent types (LlmAgent / WorkflowAgent / CustomAgent), built-in tools, MCP support, multi-agent composition, Agent Runtime sessions and memory bank, the Agent Identity SPIFFE-based auth model, and the A2A interop protocol.
- VAI.3 Vertex AI Agents — your first agent sourced
End-to-end first ADK agent — install google-adk, define a tool (Python function with type hints + docstring), define an Agent, run locally with adk run / adk web, then deploy to Agent Runtime with one Python call. Plus the Agent Identity-secured deploy variant.
- VAI.4 Vertex AI Agents vs Gemini API direct sourced
When to use Vertex AI Agents (the production agent platform on Google Cloud) vs Gemini API direct (ai.google.dev with an API key). Same SDK, different client init, very different operational footprint. Decision matrix, code difference, migration path.