Google Cloud Next 2026: Agent Platform, Vertex AI Consolidation, and Workspace Intelligence

Vertex AI absorbed into Gemini Enterprise Agent Platform — all future services exclusive to new platform

Google confirmed that the Agent Platform is 'the evolution of Vertex AI' and that all Vertex AI services and roadmap evolutions will be delivered exclusively through the Agent Platform going forward. This is a full platform consolidation — existing Vertex AI customers will migrate into the new structure with no standalone Vertex AI roadmap.

Agent Platform Build layer: Agent Studio (low-code) and Agent Development Kit (ADK, code-first)

Agent Studio provides a visual low-code interface for building agents without deep engineering. ADK is the code-first path for developers requiring full control. Burns & McDonnell is using ADK to convert project data into real-time actionable intelligence.

Agent Runtime and Memory Bank form the Scale layer — multi-day agent state and persistent long-term context

Agent Runtime enables long-running agents that maintain state for days in cloud sandboxes. Memory Bank adds persistent long-term context so agents can carry knowledge across sessions, moving agentic workloads from ephemeral to durable.

Govern layer: Agent Identity, Agent Registry, and Agent Gateway provide enterprise-grade guardrails

Each agent gets a trackable identity, is catalogued in the Agent Registry, and passes through the Agent Gateway which enforces enterprise guardrails. This directly addresses enterprise concerns around agent sprawl, auditability, and access control.

Optimize layer: Agent Simulation, Evaluation, and Observability for full execution tracing

Agent Observability provides full execution traces and real-time visibility into agent reasoning. Agent Simulation and Evaluation allow teams to test and score agent behavior before deployment — the MLOps discipline applied to agentic systems.

Model Garden offers 200+ models including Gemini 3.1 Pro, Gemma 4, and Anthropic Claude Opus/Sonnet/Haiku

The multi-model approach means enterprises are not locked into Google's own models. Anthropic's full Claude tier (Opus, Sonnet, Haiku) is available alongside Gemini 3.1 Pro, Gemini 3.1 Flash Image, Lyria 3, and Gemma 4.

Workspace Intelligence: semantic layer mapping emails, chats, files, collaborators, and projects into unified context

Workspace Intelligence operates at the Workspace application layer, distinct from the Enterprise Agent Platform infrastructure layer. It creates shared context that Gemini-powered agents use across apps, shifting Workspace from isolated tools into an interconnected intelligent system.

Ask Gemini in Chat: 'command line for work' — briefings, file retrieval, document generation, scheduling

Positioning Chat as a command line signals Google's intent to make Gemini in Chat the primary interface for knowledge work orchestration. Users can retrieve files by description, generate documents and presentations, and schedule meetings from a single conversational interface.

Sheets: natural-language spreadsheet building, HubSpot/Salesforce imports, canvas layer for dashboards and kanban

Third-party CRM data imports directly into Sheets via natural language reduce the need for manual data wrangling. The canvas layer adds dashboards, heat maps, and kanban views — bringing Sheets closer to BI tools.

Docs: AI infographic generation from business data, batch image editing, comment triage

Infographic generation grounded in business data means Docs can produce visual content tied to live or imported data. Batch image editing and comment triage address real collaboration friction in document-heavy workflows.

Slides: full editable decks in one pass using company templates and visual rules

One-pass deck generation using company-specific templates and visual rules is a significant upgrade from generic AI slide tools, keeping content within the enterprise's brand and compliance guardrails.

Drive Projects: shared context hub; Workspace MCP Server connects outside AI apps to Workspace data

Drive Projects (distinct from Gemini Enterprise's 'Projects' shared chatbot feature) acts as a contextual anchor for team initiatives. The Workspace MCP Server is the interoperability layer allowing third-party AI agents to access Workspace data — some MCP features remain in preview or private preview.

Sovereign data controls for US and EU now; Germany and India listed as future expansion

Client-side encryption, admin controls, and regional data sovereignty are positioned at regulated industries — financial services, healthcare, government — and EU customers with strict data residency requirements.

Gemini Enterprise pricing: $30/user/month (large orgs), $21/user/month (smaller orgs)

The tiered pricing creates a lower SMB entry point while maintaining premium positioning for large enterprise. Google's lower tier for smaller orgs may accelerate adoption outside the Fortune 500.

Workspace Studio GA December 2025; Personal Intelligence January 2026 preceded Cloud Next announcements

Workspace Studio GA established the agentic authoring foundation; Personal Intelligence brought Gemini deeper into the consumer app. The Cloud Next 2026 announcements complete the stack at the enterprise and platform layer.