Bolette Kern, Umbraco CMS in the Age of AI

Umbraco CMS Product Manager, Bolette Kern, outlines the strategic evolution of content management systems in the era of generative artificial intelligence. Rather than hard-coding trendy, rigid AI features directly into the core platform, Umbraco's architecture focuses on establishing an extensible framework built on choice, governance, and structural flexibility. This session breaks down how organizations can implement AI workflows on their own terms, utilizing localized control profiles, pre-generation data screening, and automated performance testing. It also highlights how the platform is preparing for an ecosystem where external AI agents read, write, and manage content alongside human creators. This video is for digital agency architects, enterprise development teams, content strategists, and product owners who need to design future-proof content platforms. It provides a technical blueprint for maintaining data privacy, evaluating non-deterministic LLM outputs, and managing concurrent multi-author content changes safely. What You Will Learn: The structural risks of vendor-locked AI strategies and why platform flexibility acts as a long-term competitive advantage. How to configure pre-generation PII guardrails using regular expressions to protect sensitive user data before it reaches external LLMs. Methods for quality-checking non-deterministic prompt outputs using automated grading criteria and side-by-side model comparisons. The architecture of Model Context Protocol (MCP) servers and open-source agent skills designed to guide AI assistants in writing idiomatic Umbraco code. A first look at upcoming core CMS features for real-time editor collaboration, property-level merge resolution, and visual text differentiation within block editors. Chapters 00:00 - The Role of a CMS in an Unpredictable Future 02:04 - Umbraco's Two-Track AI Strategy Overview 03:02 - The Risks of Hard-Coding AI into the CMS Core 05:15 - Demo: Managing LLM Connections, Profiles, and Contexts 06:14 - Enforcing PII Guardrails and Content Integrity Rules 08:36 - Testing and Grading Non-Deterministic AI Prompt Performance 13:32 - Leveraging the Umbraco AI Foundation for Custom Packages 15:30 - Track 2: Interfacing with External AI Agents 17:43 - Open-Source Agent Skills for Best-Practice Code Generation 22:46 - Model Context Protocol (MCP) Platform Architecture 26:48 - Demo: Content Orchestration via the Hosted CMS Editor MCP 31:43 - Automating Content Audits with Global Updates 35:18 - Core Platform Advantages in a Code-Generation Ecosystem 38:41 - Roadmap: Property-Level Editor Collaboration & Conflict Resolution 41:44 - Integrating Visual Diff Views into the Publishing Flow Links & Resources Umbraco, the agent-ready CMS: https://umbraco.com/products/ai/the-a... Codegarden: https://codegarden.umbraco.com/