Google Cloud unveiled the Gemini Enterprise Agent Platform at Google Cloud Next 2026 in Las Vegas. The product is explicitly positioned as the successor to Vertex AI — from here forward, all Vertex AI services and roadmap items ship through the Agent Platform rather than as a standalone product.
The platform consolidates agent development, multi-day workflow orchestration, governance, and optimization into a single surface. The core additions include a low-code Agent Studio for building agents without deep ML expertise, an upgraded graph-based Agent Development Kit for orchestrating networks of sub-agents, and the Gemini Enterprise App as the end-user surface for enterprise-wide agent discovery and deployment.
Model access is deliberately broad. The platform ships with first-class access to 200+ models including Google's own Gemini 3.1 Pro, Nano Banana 2, and Lyria 3, plus third-party models from Anthropic (Claude Opus, Sonnet, and Haiku), Gemma 4, and others. Google is betting that enterprises want optionality at the model layer while keeping orchestration, identity, and governance on one stack.
The security layer is the more distinctive piece. Agent Identity assigns every agent a unique cryptographic ID for audit and traceability. Agent Gateway acts as an 'air traffic control' layer, mediating interactions between agents and data sources. Model Armor provides specific protection against prompt injection, tool poisoning, and sensitive data leakage — a direct response to the agentic security problems that 2025 made impossible to ignore.
Announced the same week as OpenAI's Workspace Agents and against a backdrop of Microsoft's continued Copilot Agents push, the launch marks the moment when agent platforms became the primary product category of enterprise AI — not models, not chat apps, not copilots.