News
Research Report
The State of the Digital Workplace
Google launched the Gemini Enterprise Agent Platform — a single environment for enterprises to build, scale, govern and optimize AI agents — during the first day of its Google Cloud Next conference in Las Vegas. The platform brings together Vertex AI's model selection and agent-building capabilities with new features for agent integration, DevOps, orchestration and security. All future Vertex AI services and updates will be delivered exclusively through Agent Platform.
Through Model Garden, the platform provides access to more than 200 models, including first-party options such as Gemini 3.1 Pro and Gemma 4, alongside third-party models from Anthropic — including Claude Opus, Sonnet and Haiku. Agents are delivered to employees through the Gemini Enterprise app, integrated with IT operations for AI governance at scale.
The 4 Pillars of Gemini Enterprise Agent Platform
| Pillar | Description |
|---|---|
| Build | Tools to create agents, from low-code visual builder to full code development |
| Scale | Infrastructure for high-performance, long-running agents with persistent memory |
| Govern | Centralized identity, access control and security across every agent |
| Optimize | Testing, monitoring and automated refinement to keep agents on target |
Build
Build gives developers and business users flexible options depending on their technical level. Agent Studio offers a visual, low-code interface for getting started quickly, while the Agent Development Kit (ADK) supports full-code customization for complex use cases. Pre-built agent templates in Agent Garden — covering areas like financial analysis, invoice processing and code modernization — let teams deploy faster by starting from a working foundation rather than scratch.
Scale
Scale handles the infrastructure needed to move from pilot to production. The revamped Agent Runtime delivers sub-second startup times and supports agents that run continuously for days, managing multi-step workflows without losing context. Agent Memory Bank gives agents persistent, long-term memory across interactions, so they can recall user preferences and history rather than starting fresh each session.
Burns & McDonnell uses Agent Platform to transform how organizational knowledge is applied across the enterprise. Using ADK, we are building an AI agent that turns decades of project data into real-time, actionable intelligence. Agent Platform enables this innovation to scale responsibly by combining deterministic business rules with probabilistic reasoning — making AI a trusted operational capability, not just a productivity tool.
- Matt Olson, Chief Innovation Officer
Burns & McDonnell
Govern
Govern establishes centralized control over every agent in an organization's fleet, whether built in-house or sourced from a partner. Each agent is assigned a unique cryptographic identity, creating an auditable trail for every action it takes. Agent Gateway acts as a central control point, enforcing security policies and protecting against threats like prompt injection and data leakage. A unified Agent Security Dashboard, powered by Google's Security Command Center, provides real-time threat detection and vulnerability scanning.
Optimize
Optimize ensures agents perform reliably once deployed. Agent Simulation lets teams test agents against realistic synthetic conversations before launch, automatically scoring them on task success and safety. In production, Agent Evaluation continuously monitors live performance across full conversations, while Agent Optimizer analyzes real-world failures and suggests improvements automatically.
Google's AI Strategy in Context
Google has made AI integration the centerpiece of its enterprise strategy over the past year. It opened Workspace to third-party models in April 2025, including Anthropic's Claude. That same month, Google launched the Agent2Agent (A2A) protocol, an open standard for cross-vendor AI agent coordination. The protocol transferred to the Linux Foundation in June 2025 and drew more than 150 organizational supporters by July, including a formal Microsoft commitment in September.
The company introduced Google Enterprise in October 2025, a unified platform connecting AI models, productivity tools and integrations with Salesforce and SAP, laying the organizational groundwork that Agent Platform now builds on.
On pricing, Google bundled Gemini into Workspace plans in January 2025, raising base prices and retiring the standalone add-on. Earlier this month, the company moved previously free AI features behind a paywall.
Have a tip to share with our editorial team? Drop us a line: