AI, intentionally
Umbraco is not a one-size-fits-all CMS.
It’s a flexible content platform used in countless different ways, across industries, organisations, and workflows. Introducing AI into such an adaptable product isn’t just about what AI can do. It’s about who it serves, how it behaves, and whether it still makes sense a year from now.
We’ve listened. We’ve experimented. And we watched how AI evolved, because it is evolving fast.
Today, we’re ready to share the next step in Umbraco’s AI journey. One that respects Umbraco as a product, supports our partners, elevates our community, and gives users a real choice, without locking them into decisions they’ll regret later.
A platform built for choice
Our answer isn’t to launch a big AI feature and call it done.
Instead, we are doubling down on what Umbraco does best: being a stable, flexible platform that can adapt to many futures, not just one predicted path.
That’s why our AI strategy is built around three principles.
AI on your terms
You choose if, how, and where AI is used.
No forced models. No hidden costs. No untransparent “magic”.
Instead of outsourcing your AI strategy to the platform, Umbraco lets you define it yourself. This makes AI a safer long-term choice and a far easier conversation when it comes to governance, compliance, and trust.
Modular AI experiences
AI will change. Your platform shouldn’t have to change every time it does.
Rather than hard-coding AI into the core, we believe in AI as modular building blocks. Some built by Umbraco HQ. Others by partners. Others again by the community. Use what you need. Extend when you want. Ignore the rest.
AI everywhere Umbraco is
AI doesn’t live in one place. Developers have their tools. Editors have their workflows. Automation often lives outside the CMS entirely.
Umbraco is designed to work across all of this, inside the backoffice, alongside external tools, and as part of larger AI-driven systems.
In short:
Umbraco stays stable. AI stays flexible. You stay in control.
How Umbraco becomes a platform for AI
Umbraco’s AI strategy takes shape across two connected tracks: AI in Umbraco and Umbraco in AI. Together, they describe how AI can live both inside the platform and around it, without compromising the stability of the core product.
A solid foundation for AI in Umbraco
Introducing AI into the Umbraco backoffice requires more than adding features. It requires a foundation that teams can trust, today and over time.
That foundation is Umbraco.AI
Umbraco.AI is an open-source package, owned and maintained by Umbraco HQ. This is a deliberate choice. While innovation should come from many places, the core AI foundation inside Umbraco should be stable, well-defined, and looked after over time. That way, partners and end customers don’t have to worry about long-term maintenance or direction.
Umbraco.AI is where you decide how AI works inside your Umbraco backoffice.
This is where you configure:
Which AI providers and models you want to use - yes, plural. Umbraco doesn’t decide which provider or model you choose, or how many you switch between. That’s completely up to your organisation and your needs.
How different models are used, for what purpose, and where AI is allowed to operate.
Tone, behaviour, and brand alignment
Logging, auditing, and traceability
Versioning and rollback, so changes can be reviewed and reversed
In short, this is where AI behaviour is defined, before it’s ever surfaced to users.
On top of this foundation is a modular AI layer. This is where AI capabilities add value, through experiences like Umbraco Copilot, prompt, automation, and future extensions. These packages can be built by Umbraco HQ, by partners, or by the community, but they all rely on the same Umbraco.AI configuration.
In practice, that means whatever AI package you choose to adopt, they respect the same rules you’ve already defined in Umbraco.AI. Governance is defined once and consistently applied across everything that builds on top of it.
We’re not shipping AI as one tightly coupled, one-size-fits-all capability. We’re shipping it as modular building blocks you can pick from - adopt, extend, or ignore.
A strong base, then build as you want to build
With the Umbraco.AI release, we can’t wait to get your input on this foundation package for future AI functionality. Umbraco.AI is released together with five connections to AI providers (Anthropic, OpenAI, Amazon, Google, and Microsoft Foundry. This gives teams immediate choice, both in providers and models, right from the start.
We’re also looking to build packages for the modular AI layer. We’re first looking into broadly useful capabilities: AI features that most websites and editorial teams will recognise and request, regardless of industry and scale.
Umbraco.AI.Prompt introduces predefined, predictable AI actions directly in the Umbraco Backoffice. You define the action in a prompt and control where it can be used, on which document types and fields. Editors can then use these actions on content and media, for example, generating alt text, SEO titles, summaries, or tag content, all in a way that is consistent, safe and easy to understand. You can also find Umbraco.AI.Prompt on Umbraco Marketplace.
Umbraco.AI.Agent & Copilot bring a familiar copilot experience into the Umbraco Backoffice. Editors get a chat interface that understands where they are in the application and what they are working on. The copilot can suggest changes, help clean up content, assist with structure, compare and analyse content, etc. All while always keeping the human in the loop.
We’re also exploring other ways AI can support editors in their everyday work, especially where repetitive tasks, consistency issues, and content maintenance tend to slow teams down. This could include anything from helping keep content consistent and well-structured, to supporting reuse across channels and managing content throughout its lifecycle. More on this later.
Your clients, your ideas, your AI solutions
Alongside what we build at Umbraco HQ, we see enormous potential in the innovation coming from partners and the community. Umbraco.AI, Umbraco Prompt, Umbraco Agent, and Umbraco Copilot are shipped as open source. That openness is what allows AI in Umbraco to grow in many directions, from broadly useful packages to more specialised solutions.
We invite partners and the community to take this foundation and build on it: custom marketing flows, industry-specific assistants, advanced automation, and workflow tools tailored to real client needs.
For partners, this is an opportunity to retain the role of the expert, not just the installer. Instead of just configuring the same out-of-the-box AI features as everyone else, partners can build unique AI capabilities that reflect their expertise and their client’s business. This could be predictive search and intent modeling, intelligent asset tagging, advanced image editing, super-personalised content recommendation, dynamic A/B test and optimization - the real differentiation for your clients.
AI Community intiatives
Like with everything else we do, we want to work closely with the partners and community members who are pushing AI forward. That’s why we’re creating two new AI Community initiatives, working alongside us at Umbraco HQ and feeding real-world experience directly into the product. Applications are open now.
Learn and read more about the AI in Umbraco Advisory Board, or explore the Umbraco in AI community team.
And AI creativity isn’t limited to formal teams. From meetups to hackathons and events like Umbraco Spark this March, where both the Hackathon and Package Jam focus on AI, we can’t wait to see what the ecosystem builds next.
Umbraco in external AI tools
AI doesn’t live in one product or one place, and Umbraco is built to work wherever AI happens.
With Umbraco in AI, we gather all the tools from Umbraco HQ that make Umbraco work naturally together with external AI systems. Here, Umbraco acts as the trusted source of content and structures that AI tools can interact with, whether that’s a coding assistant, an automation flow, or an agent-based setup.
We took the first step in this direction with Umbraco Developer MCP in June 2025, enabling AI tools to communicate with Umbraco through the Management API. Since then, continued to work on what we call agent resources. Tools that help AI understand how to work with Umbraco in a reliable and intentional way.
One of these is a base implementation for MCP. Think of it as the Umbraco recipe for your external AI setup. It provides a set of tools and a template to help you build your own MCP implementations based on Umbraco. For partners, this means external AI tools can work more accurately with Umbraco, without having to rediscover the same patterns again and again. The Umbraco base implementation will be released as a beta soon.
We’re also introducing Agent Skills, which help guide LLM away from purely probabilistic answers and towards information you actually want it to use. The Agent Skills are integrated with the latest Umbraco documentation and also integrate with examples of best practise, or specific guidance, so AI responses are not just plausible, but relevant and correct.
Together, these tools make Umbraco easier to implement in AI-driven setups outside the Umbraco products, without trying to own or control those environments. Umbraco stays open, predictable, and well-defined, even when it’s being used by systems beyond Umbraco.
AI, the Umbraco way
AI isn’t a feature. And it’s definitely not finished.
Our goal isn’t to predict exactly how everyone will use AI. It’s to build a platform that still works when those predictions change. We’re not optimising for the loudest AI features. We’re optimising for choice, longevity, and real-world use. Because when the hype moves on, and AI becomes everyday infrastructure, that’s what partners and end clients will rely on.
We’re excited about what we’re building, but the real impact comes from what the Umbraco ecosystem builds next.
Quick note: We also explored Umbraco’s AI strategy at the Umbraco Winter Keynote 2026 earlier this week - watch the recording here.