_homepageheader.webp)
Using the AI ecosystem inside and outside the CMS to refine content and even create entire websites: is that possible?
Let’s imagine this: we’re working in a Content Management System (CMS) and want to ensure that the text we’re editing is aligned with market trends.
It should reference what sells well, what’s being talked about, or what promises high growth rates – our goal is to increase the time, visitors stay on our site, their engagement, and conversion rates.
To achieve this, we could conduct a thorough analysis and research which products, topics, or companies are trending, and invest in weaving these into the text.
This would likely be a mix of manual work and using tools specialized in tracking market trends – including some powered by AI.
Usually, all of this happens outside the CMS, since its main task isn’t to track the market or optimize content against trends. We’d have to leave the CMS, switch between different tabs and tools, rewrite text or have it rewritten, and finally hit “publish” from within the CMS.
Imagine this: there’s an easier way.
And that is through a connection between the AI ecosystem – which can also perform market analysis – and the CMS.
That very connection is called: Model Context Protocol (MCP).
This protocol is a standard that enables AI to access information and tools it otherwise wouldn’t have. For example, market trends. But also customer, product, or even weather data, tickets in Jira, designs in Figma, or shared notes in Confluence.
Through MCP, AI – more precisely: the large language model – can access all of this.
It’s no coincidence MCP is often compared to USB-C ( see here ); a universal interface through which i.e. a text in the CMS can be optimized against market trends – assuming the CMS supports MCP.
Here’s how it works, technically speaking: as an “MCP Client,” the CMS sends the request “optimize text based on market trends” to an “MCP Host.” The host then connects to a suitable “MCP Server” – let’s call it the “AI-powered trend analysis” tool – where the request is processed.
If we’re writing a text about summer fashion, the AI might highlight flared trousers. Or white peaches, if we’re writing about trending fruit salad ingredients. Or it could be the hybrid headless CMS that boosts efficiency and scalability with the AI Suite.
A transparent CMS marks these suggestions so users can decide what to include in the text. Only once we’re satisfied with the AI suggestions do we hit publish.
This shows how MCP can be used to access tools in the AI ecosystem from within the CMS.
But it also works the other way around.
Accessing the CMS with the AI tool – from “outside” to “inside.”
For example, if we want to create a website without being inside the CMS.
Sounds crazy?
It isn’t: we can use AI tools like Claude or ChatGPT to send the request “create a website about a modern hybrid headless CMS that boosts efficiency and scalability with the AI Suite” to the CMS.
Of course, the CMS must be MCP-compatible for that to work.
It must have the necessary actions and elements defined so the AI can access and populate them, for example: This is a header; it needs a headline of no more than 10 words. Or: This is a product description that should follow the company’s tone of voice. Or: This is an image that needs a description reflecting the page context to improve accessibility.
When these requirements are met, we see the website come to life before our eyes – with page structure, content, and text placed meaningfully in the right template areas – all without us doing more than submitting the initial request or being CMS experts.
And here too, a transparent AI-enabled CMS allows users to decide whether the page is published immediately or first reviewed by humans, who only hit publish once they are satisfied with the AI’s suggestions.
If you don’t just want to imagine all this, but actually experience it live: FirstSpirit shows how AI can already be used today – efficiently, compliantly, and at scale. Feel free to get in touch!
Timo Klattenhoff
Head of Product
FirstSpirit – A Crownpeak Solution
Sources