Gemini Enterprise Agent Platform: the new era of enterprise AI agents
Google has just launched Gemini Enterprise Agent Platform, the evolution of Vertex AI . It is a comprehensive platform for building, scaling, governing, and optimizing AI agents in production. We explain what is changing and what this means for your company.
In the early days of generative AI, building reliable business tools required an enormous engineering effort. Vertex AI resolved much of that complexity. But the landscape has changed: we now manage agents that interact with multiple systems—often without the necessary security and governance layers.
Google has just taken the next step. As part of the Google Cloud Next '26, has launched Gemini Enterprise Agent Platform: a comprehensive platform for building, scaling, governing, and optimizing AI agents that act with the same reliability as any other member of your team.
The most important thing for your company: Vertex AI isn't going away—it’s evolving. All the model selection, building, and training capabilities you were already using are now integrated into Agent Platform, alongside new features for agent identity, DevOps, orchestration, and security. If you already work with Vertex AI , the transition is seamless.
As Google Cloud Premier PartnerAt The Cloud Collective , we are ready to help you leverage this platform from day one. Below, we explain what it includes, what changes, and how it can transform your operations.
The 4 pillars of Agent Platform
Agent Platform organizes the entire agent lifecycle into four pillars, covering everything from development to optimization in production.
Building agents
Choose from the new one Agent Studio (visual, low-code) or the Agent Development Kit (ADK) enhanced (code-first). Both enable moving from idea to production in hours, with AI-assisted coding capabilities.
Scale agents
The new one Agent Runtime enables cold starts of less than a second and agents that run autonomously for days, maintaining state with Memory Bank for long-term persistent context.
Governing agents
Each agent receives one. unique cryptographic identity with Agent Identity, a centralized registry via Agent Registry, and traffic control via Agent Gateway. enterprise -grade security by design.
Optimize agents
Simulate realistic conversations before launch, monitor in production with full traces of the agent's reasoning, and let Agent Optimizer automatically suggest improvements.
Access to over 200 models from Model Garden
Agent Platform maintains premium access to the broadest model catalog on the market, including Google's latest first-party releases and third-party models.
Gemini 3.1 Pro
Google's most advanced model for complex reasoning and multimodal tasks.
Gemini 3.1 Flash Image
Native image generation and editing with optimized speed.
Gemma 4
Google's open-source model for on-premise and edge deployments.
Lyria 3
Google's new model for high-quality audio and music generation.
Claude ( Anthropic )
Opus, Sonnet, and Haiku available as third-party models within the platform.
Over 200 models
Total flexibility to choose the best model for each use case.
The most significant technical developments
What changes with Agent Platform
🔄 Vertex AI → Agent Platform
All Vertex AI services and the roadmap are now delivered exclusively through Agent Platform. It is not a separate product; it is the natural evolution.
🧠 Memory Bank
Agents generate and curate long-term memories of conversations. With Memory Profiles, they recall details with high precision and low latency.
🔐 Agent Identity
Each agent receives a unique cryptographic ID. Every action is audited and mapped to authorization policies defined by your team.
⚡ Revamped Agent Runtime
Sub-second cold starts, multi-day workflows, and agent-to-agent orchestration using deterministic and generative patterns.
🛡️ Agent Gateway
Centralized traffic control between agents and tools across any environment, with integrated Model Armor against prompt injection and data leaks.
🔬 Agent Optimizer
It automatically analyzes production failures, groups them by pattern, and suggests improved instructions to correct the agent's behavior.
How global companies are using it
These are some of the companies already transforming their operations with Agent Platform:
They use ADK to convert decades of project data into actionable intelligence in real time. The agents combine deterministic business rules with probabilistic reasoning.
Its Color Assistant, built with ADK and Agent Runtime, automates breast cancer screening: it verifies eligibility, connects users with clinicians, and schedules appointments at scale.
They rebuilt their Xfinity Assistant using ADK and a multi-agent architecture, moving from basic scripts to generative conversational support with first-interaction resolution.
They built their own Beauty Tech Agentic Platform on Google Cloud . They use ADK for the autonomous orchestration of agents connected to their data sources via MCP.
What does this mean for your company in Spain?
Agent Platform shifts the paradigm: it is no longer about managing individual AI tasks, but rather about delegate end-to-end business outcomes to agents with identity, memory, and enterprise governance.
If you already use Vertex AI
The transition is seamless. Everything you currently have continues to work, and you now have access to the new governance layers (Agent Identity, Registry, Gateway) and scaling capabilities (Memory Bank, revamped Runtime) without needing to migrate anything.
If you haven't started with AI yet
Agent Platform is the best possible entry point. You can start with Agent Studio (visual, no-code) and scale up to ADK when you need more control. Pre-trained models from Agent Garden give you an immediate head start for use cases such as invoice processing, financial analysis, or code modernization.
If you are concerned about security and governance
It is the first agent platform that treats security as a fundamental pillar, not as an add-on. Agent Identity, Model Armor, Agent Anomaly Detection, and integration with Security Command Center provide the level of control European companies need to comply with the GDPR and the AI Act.
Do you want to implement Agent Platform in your company?
As a Google Cloud Premier Partner , we help you design, build, and deploy AI agents using Agent Platform from day one. Identity, governance, and scaling included.
Speak with our teamFrequently Asked Questions
No. Vertex AI is evolving and fully integrating into the Gemini Enterprise Agent Platform. All existing capabilities for model selection, training, and agent building remain available, now with new governance, identity, and scaling features. There is no need to migrate anything; it is a natural evolution.
Agent Identity assigns a unique cryptographic identifier to each agent. Every agent action is logged and mapped to authorization policies. This is critical for regulatory compliance ( GDPR , AI Act) and for maintaining full traceability of each agent's activities within your organization.
Not necessarily. Agent Studio offers a low-code visual interface for building agents without writing code. When you need more customization, you can export the logic to the ADK (Agent Development Kit) and continue in a full-code environment. There are options for all technical profiles.
Over 200 models are available in Model Garden, including Gemini 3.1 Pro , Gemini 3.1 Flash Image, Gemma 4, Lyria 3, and third-party models such as Anthropic 's Claude . You can choose the best model for each task without switching platforms.
Yes. Agent Platform includes security by design with Agent Identity, Agent Gateway with Model Armor (anti-prompt injection), Agent Anomaly Detection, and integration with Security Command Center. Additionally, Google Cloud offers data residency in the EU, is GDPR compliant, and holds ENS High-Level validation.
As a Google Cloud Premier Partner , we help you design agent architecture, select the right models, implement governance and security layers, and train your team to operate autonomously—from proof of concept to production, with dedicated support.
