We Moved Our Content Out of the CMS So AI Agents Could Actually Read It
Bodil Biering
A CTO told us this in a recent AI builders club call:
We recently removed our bible. All of our content — pages, structure, links — moved out of the CMS and into Markdown files, directly in the code.
The reason wasn’t UX, and it wasn’t a technical preference. It was about AI agents.
If an agent is going to look at what’s currently there, understand how things relate, decide on structure, or touch links safely, it needs to actually be able to read the content. Not fetch it from an external CMS. Not chase down images sitting in a third-party CDN. The moment content lives outside the codebase, an agent’s ability to form good context — and make a good decision — gets shaky.
Once you accept that agents should be able to work with this material directly, keeping content external stops making sense.
The insight came from a CTO, who wanted to empower everyone on the team to use AI agents safely — meaning with proper version control, and alignment between all the pieces of information a decision depends on. A CMS and a CDN sitting outside your codebase can’t offer either.
So this wasn’t a redesign. It was an infrastructure decision about what safe AI use actually requires.
