--- title: "From CMS to AI agent — how we reinvented our website — Sping Digital Lab" source: "https://sping.nl/en/insights/from-cms-to-ai-agent/" --- 1. [Home ](https://sping.nl/en/)/ 2. [Blog ](https://sping.nl/en/insights/)/ 3. From CMS to AI agent — how we reinvented our website 3 minutes to read # From CMS to AI agent — how we reinvented our website For years, Sping ran its website on a CMS. Practical, but limiting. We dropped it — and discovered that an AI agent is a better content manager than we ever were ourselves. ![For years, Sping ran its website on a CMS. Practical, but limiting. We dropped it — and discovered that an AI agent is a better content manager than we ever were ourselves.](https://sping.nl/images/posts/van-cms-naar-ai-agent/00-cms-ai-agent.png) For years it was a given: every website needs a CMS. The presentation layer separated from the content. Content managers could update copy without breaking anything. Developers were needed for bigger changes. On the surface, a clean division of responsibilities. In practice, it worked differently. ## WordPress: flexible — until it wasn’t We started with WordPress. It worked well at first. But as requirements grew, so did the plugin list. Conflicts, updates, slowdowns. What started as a manageable system gradually became a millstone. Everything was entangled, and every update could break something else. ## Umbraco: more stable, but still in the way We moved to Umbraco. Better technical foundation — more stable, easier to maintain. But it got in our way whenever we wanted to change something quickly. Any change beyond replacing text meant a round trip through a developer. For a small, driven team with fast-changing ideas, that friction adds up. ## The real problem: we were doing everything ourselves But the biggest issue wasn’t the CMS itself. It was that every single change had to be made by our own team. A lot of clicking, copying and pasting, dragging and rearranging. Time-consuming manual work. For other projects, we’d already left this behind. We’d been using AI-assisted development: tools where the AI has full context of the project and generates new code based on a vast amount of training data. It worked so well that we started asking: why aren’t we doing this for our own website? ## Content and code in the same codebase That insight sparked an idea: drop the CMS. Choose a framework where content and presentation live in the same codebase. The AI agent can see both and use them together to write, restructure, and publish new content. We chose Astro — a modern static framework where Markdown files sit right next to the code. No separate database, no query layer, no CMS interface. During the build process, content is compiled into static pages. That means no computing power needed at request time. No queries. Blazing fast. ## The new workflow: describe instead of click ![From description to Pull Request, preview, and live](https://sping.nl/images/posts/van-cms-naar-ai-agent/01-cms-ai-agent-2.png) The way we work is fundamentally different now. We describe what we want, much like we used to ask a content manager. But now we ask an AI agent. The agent understands the structure of the website, knows the existing content, and knows how everything fits the style and brand. Based on our description, it prepares a Pull Request containing all the changes — from new copy to layout adjustments. From that PR, a preview version of the website is automatically built. We review the preview, read through the code, and give feedback if something’s off. Once we’re happy, we merge to the main branch — and the change goes live. ## Including this post This article itself was made this way. We described the story, the AI wrote the text, and then we had a Claude Skill built to make this process even smoother next time — with high consistency and less friction. Over the past few years, we published very few blog posts. The knowledge and experience were always there — we just thought we didn’t have the time alongside all the projects. Fortunately, that’s changed now, as this post proves. We’re committed to making and keeping our website current and relevant. _(Disclaimer: we used AI to write and refine this article. The content is based on our own experiences and was provided by us.)_ * [Artificial Intelligence](https://sping.nl/en/tag/artificial-intelligence/) * [Agentic Ai](https://sping.nl/en/tag/agentic-ai/) * [Static Site](https://sping.nl/en/tag/static-site/) * [Cms](https://sping.nl/en/tag/cms/) * [Astro](https://sping.nl/en/tag/astro/) Share: [Back to blog](https://sping.nl/en/insights/) ## Related posts [View all posts »](https://sping.nl/en/insights/) [![How we learned to trust our AI assistant](https://sping.nl/images/posts/hoe-we-leerden-onze-ai-assistent-te-vertrouwen/00-blog-ai-assistent-vertrouwen.jpg)](https://sping.nl/en/insights/how-we-learned-to-trust-our-ai-assistant/) ### [How we learned to trust our AI assistant](https://sping.nl/en/insights/how-we-learned-to-trust-our-ai-assistant/) Sping built a sales assistant that challenged our assumptions about AI autonomy — and revealed that trust in algorithms works much like trust in human colleagues: through context and dialogue, not rigid rules. [![How we built a voice assistant that actually delivers](https://sping.nl/images/posts/hoe-we-een-voice-assistant-maakten-die-echt-iets-kan/00-blog-voice-assistant-maken.jpg)](https://sping.nl/en/insights/how-we-built-a-voice-assistant-that-actually-delivers/) ### [How we built a voice assistant that actually delivers](https://sping.nl/en/insights/how-we-built-a-voice-assistant-that-actually-delivers/) Sping combined OpenAI's Realtime API with WebRTC and HubSpot CRM into a voice assistant that executes complex CRM actions faster than humans could perform them manually. [![Sping builds tour guide with artificial intelligence](https://sping.nl/images/posts/highlight-delft-sping-bouwt-tourguide-met-kunstmatige-intelligentie/00-1-skyggen-aveny-t-foto-carl-emil-carlsen.webp)](https://sping.nl/en/insights/highlight-delft-sping-builds-tour-guide-with-artificial-intelligence/) ### [Sping builds tour guide with artificial intelligence](https://sping.nl/en/insights/highlight-delft-sping-builds-tour-guide-with-artificial-intelligence/) Sping is developing an AI-powered chatbot tour guide for Highlight Delft 2024, using large language models to provide personalised answers to visitor questions about artworks. [![Prompt-engineering for GPT-4 and other language models](https://sping.nl/images/posts/highlight-delft-prompt-engineering-voor-gpt-4-en-andere-taalmodellen/00-w661_1371-zeno-van-den-broek-forming-folds-700x467.jpg)](https://sping.nl/en/insights/highlight-delft-prompt-engineering-for-gpt-4-other-language-models/) ### [Prompt-engineering for GPT-4 and other language models](https://sping.nl/en/insights/highlight-delft-prompt-engineering-for-gpt-4-other-language-models/) Developer Kris explains how Sping uses AI to power the Highlight Delft 2024 tour guide, and shares six essential tips for writing effective prompts for language models.