Google Gemini enterprise agent platform

Published April 27, 2026 by The Daily Scout

- Google Cloud on April 22 unveiled Gemini Enterprise Agent Platform, recasting Vertex AI as its main system for building, deploying, governing and optimizing enterprise AI agents. - Google said the platform’s reworked runtime supports agents that keep state for days, with Memory Bank for persistent context and access to 200-plus models. - The move folds Google’s AI tools into one enterprise stack as rivals push agent platforms and model choice. (cloud.google.com)

Why it matters

Google Cloud has turned Vertex AI into Gemini Enterprise Agent Platform, a new system for building and running enterprise AI agents. (cloud.google.com) Google announced the change on April 22 at Cloud Next 2026 and said future Vertex AI services and roadmap updates will ship through Agent Platform instead of as a standalone service. (cloud.google.com) The company said the platform combines model selection, model building and agent building with new tools for integration, DevOps, orchestration and security. (cloud.google.com) In plain terms, the pitch is that companies want software that does more than answer prompts. They want agents that can call tools, move across systems, remember prior work and keep running after a single chat ends. (cloud.google.com) (docs.cloud.google.com) Google’s key technical claim is about duration and memory. It said the re-engineered Agent Runtime supports long-running agents that maintain state for days and uses Memory Bank for persistent, long-term context. (cloud.google.com) (docs.cloud.google.com) Memory Bank stores information from conversations across multiple sessions, then retrieves it later so an agent can personalize responses and keep continuity without stuffing everything into one context window. (docs.cloud.google.com) Google is also emphasizing model choice, not just Gemini. Agent Platform says it offers access to more than 200 models in Model Garden, including Google models, open models such as Gemma and Llama, and Anthropic’s Claude family. (cloud.google.com 1) (cloud.google.com 2) The platform includes a low-code Agent Studio, a code-first Agent Development Kit, retrieval-augmented generation tools for connecting private data, evaluation tools and tracing for debugging tool calls and reasoning loops. (docs.cloud.google.com) (cloud.google.com) Google is packaging the governance layer alongside the developer tools. It said IT teams get a single control plane, identity and access controls through Google Cloud Identity and Access Management, policy controls and monitoring of agent traffic and outputs. (cloud.google.com) (docs.cloud.google.com) The broader Gemini Enterprise lineup now pairs this developer platform with the Gemini Enterprise app, where employees can discover, share and run agents inside the same governed environment. Google also said partners including Oracle, Salesforce and ServiceNow will distribute third-party agents through the ecosystem. (cloud.google.com) The result is a name change with a product strategy behind it: Google is moving Vertex AI from a model-and-tools brand to an agent platform brand built around longer-running software workers. (cloud.google.com)

Key numbers

Google Cloud on April 22 unveiled Gemini Enterprise Agent Platform, recasting Vertex AI as its main system for building, deploying, governing and optimizing enterprise AI agents.
Google said the platform’s reworked runtime supports agents that keep state for days, with Memory Bank for persistent context and access to 200-plus models.
(cloud.google.com) Google announced the change on April 22 at Cloud Next 2026 and said future Vertex AI services and roadmap updates will ship through Agent Platform instead of as a standalone service.
Agent Platform says it offers access to more than 200 models in Model Garden, including Google models, open models such as Gemma and Llama, and Anthropic’s Claude family.

What happens next

(cloud.google.com) Google announced the change on April 22 at Cloud Next 2026 and said future Vertex AI services and roadmap updates will ship through Agent Platform instead of as a standalone service.
Google also said partners including Oracle, Salesforce and ServiceNow will distribute third-party agents through the ecosystem.

Sources

Quick answers

What happened in Google Gemini enterprise agent platform?

Google Cloud on April 22 unveiled Gemini Enterprise Agent Platform, recasting Vertex AI as its main system for building, deploying, governing and optimizing enterprise AI agents. Google said the platform’s reworked runtime supports agents that keep state for days, with Memory Bank for persistent context and access to 200-plus models. The move folds Google’s AI tools into one enterprise stack as rivals push agent platforms and model choice. (cloud.google.com)

Why does Google Gemini enterprise agent platform matter?

Google Cloud has turned Vertex AI into Gemini Enterprise Agent Platform, a new system for building and running enterprise AI agents. (cloud.google.com) Google announced the change on April 22 at Cloud Next 2026 and said future Vertex AI services and roadmap updates will ship through Agent Platform instead of as a standalone service. (cloud.google.com) The company said the platform combines model selection, model building and agent building with new tools for integration, DevOps, orchestration and security. (cloud.google.com) In plain terms, the pitch is that companies want software that does more than answer prompts. They want agents that can call tools, move across systems, remember prior work and keep running after a single chat ends. (cloud.google.com) (docs.cloud.google.com) Google’s key technical claim is about duration and memory. It said the re-engineered Agent Runtime supports long-running agents that maintain state for days and uses Memory Bank for persistent, long-term context. (cloud.google.com) (docs.cloud.google.com) Memory Bank stores information from conversations across multiple sessions, then retrieves it later so an agent can personalize responses and keep continuity without stuffing everything into one context window. (docs.cloud.google.com) Google is also emphasizing model choice, not just Gemini. Agent Platform says it offers access to more than 200 models in Model Garden, including Google models, open models such as Gemma and Llama, and Anthropic’s Claude family. (cloud.google.com 1) (cloud.google.com 2) The platform includes a low-code Agent Studio, a code-first Agent Development Kit, retrieval-augmented generation tools for connecting private data, evaluation tools and tracing for debugging tool calls and reasoning loops. (docs.cloud.google.com) (cloud.google.com) Google is packaging the governance layer alongside the developer tools. It said IT teams get a single control plane, identity and access controls through Google Cloud Identity and Access Management, policy controls and monitoring of agent traffic and outputs. (cloud.google.com) (docs.cloud.google.com) The broader Gemini Enterprise lineup now pairs this developer platform with the Gemini Enterprise app, where employees can discover, share and run agents inside the same governed environment. Google also said partners including Oracle, Salesforce and ServiceNow will distribute third-party agents through the ecosystem. (cloud.google.com) The result is a name change with a product strategy behind it: Google is moving Vertex AI from a model-and-tools brand to an agent platform brand built around longer-running software workers. (cloud.google.com)