Agent building & orchestration
Google Cloud's renamed Vertex AI agent stack, combining a managed agent runtime with sessions, a memory bank, example store and evaluation service, and support for ADK, LangGraph, LlamaIndex and A2A agents. Governance evidence centres on registration through the Gemini Enterprise app and built-in tracing rather than approval workflows.
commercial · generally available · Research snapshot 2026-09-06
Visit the official product source ↗Agent building & orchestration · Observability & traceability · Evaluation & testing
Useful conversation with: AI platform lead, Cloud architect, ML engineering manager.
Demonstrate deploying an ADK agent to the managed runtime, registering it in Gemini Enterprise, and inspecting its traces in Cloud Trace.
Support labels reflect the supplied research. Documentation and vendor claims are not independent product tests. “Not established” means the researcher did not find support; it does not prove a capability is absent.
Documented by provider
Provides a fully managed agent runtime to deploy and scale agents without managing infrastructure, with built-in observability and multiple deployment paths.
Limit: Documentation does not state tenant isolation guarantees for the managed runtime.
Documented by provider
Includes Sessions for continuous conversation state, a Memory Bank for persistent long-term memories with revision inspection, and an Example Store plus Evaluation Service to test, monitor and trace agent behaviour.
Limit: Memory retention policies and data residency controls are not described on this page.
Documented by provider
Supports custom agents plus LangChain, LangGraph, LlamaIndex query pipeline, Agent2Agent, ADK and AG2 agents.
Limit: Feature parity across frameworks is not stated.
Vendor claim
Custom-built agents can be securely registered, managed and governed using the Gemini Enterprise app.
Limit: Product page wording; the specific governance controls (permissions, approvals, audit) are not enumerated.
Not established in this research
A built-in human approval or action-gating workflow for agent tool calls is not established by the pages reviewed.
Limit: Absence of evidence; ADK may provide developer-level patterns not covered here.
Listing does not imply partnership, supplier status, a working DutyGraph integration, or a compliance certification.
Suggest a correction