Here's a detailed two-layer AI architecture — AI embedded inside the CMS for editors, and the CMS exposed as callable tools for autonomous agents — positioning dotCMS as the governed content backbone for the agentic AI era.
Agentic AI marks a shift from assistants that suggest to agents that act: calling functions, reading structured data, and chaining operations across systems without waiting for a human click. Gartner projects that 40% of enterprise apps will feature task-specific AI agents by 2026, up from under 5% in 2025. Most CMS platforms were designed for humans navigating menus — they bolt on AI features but don't expose their functions as tools. dotCMS is designed for exactly that.
The Two-Layer Architecture
Layer 1 — AI inside dotCMS. dotAI is built into the content lifecycle: AI-assisted content and image generation, asynchronous translation at scale, vector-based semantic search, and configurable AI workflow sub-actions.
Layer 2 — dotCMS available to AI. Every function, content type, workflow, and asset is exposed as a callable tool through the dotCMS MCP Server, dotCLI, and agent-consumable REST and GraphQL APIs. The same functions an editor uses through the UI are the functions an agent calls programmatically — on the same data, governed by the same rules.
The MCP (Model Context Protocol) Server gives agents live context — content types, field definitions, workflow states, permissions, and audit state — before they act, eliminating the schema guessing and hallucinated field names that cause generic LLMs to fail against unfamiliar platforms. Current capabilities cover the full content operations lifecycle, from querying and modifying schemas to moving content from draft through publish and archive.
Governance Is the Architecture
Agents operating via the MCP Server inherit the same role-based access controls governing human editors, and every workflow action is recorded in a native audit trail. The 2026 roadmap extends this: Multi-Provider Support (Q2 2026) routes AI operations to approved model providers such as Azure OpenAI, AWS Bedrock, and Google Vertex; Brand Voice and Content Standards configuration (Q3 2026) applies a single policy ruleset to every AI output; and a Content Quality Agent plus Admin Observability and Governance (Q4 2026) make AI actions as visible and auditable as any other content operation.
Why It Matters
Models will commoditize. What will not commoditize is a content infrastructure designed to put any capable agent to work on governed data, through stable contracts, with a complete audit trail. The content types that structure your web content are the same schemas agents read; the APIs that power your headless front-ends are the same APIs agents use to build, translate, and publish at scale.
Read the full blog: https://www.dotcms.com/blog/from-ai-assisted-to-agent-ready-how-dotcms-is-building-for-the-agentic-age