Migrating to a new CMS is the step teams underestimate. See how AI reshapes it — and turns a painful move into a chance to improve your content.
It used to be that content migration was a difficult problem. Difficult for a lot of reasons: the sheer number of pages, for example, or misaligned and outdated markup. Migration was the step that everyone underestimated. We always hoped for something that could make it easier — that could intelligently automate the most frustrating parts of the process.
And now, we find ourselves in the age of AI, and migration is …
… still a difficult problem.
A difficult problem, but more importantly, an opportunity. And while AI has clearly overhauled the process of migrating content from one system to another — making it more efficient and effective — the new complication is how we use AI to open doors beyond the migration itself, improving and expanding the content itself in the process.
Here’s what migration looks like now in the age of AI.
Discovery
The first step of any migration is understanding the content to be moved. Depending on the project, this could mean anything from reading simple database schemas to browsing a site, reading markup, taking careful notes, and making educated guesses. There are a lot of existing complications, which potentially makes discovery error-prone and time-consuming.
This is one of the biggest changes. Now we have automated agents that can ingest an entire website, bucket pages based on similarity, analyze markup, and make even more educated guesses about content types, functionality, and integration points. The result is a more detailed and comprehensive view with significantly less effort.
This detailed and comprehensive view isn’t just nice to have — it drastically improves our understanding of the source content, upon which we can make smart decisions about mapping existing content onto the structure of the new site. Sometimes this is simple — articles go in the article feed, for example — while other times the restructuring requires some thought. Once again, AI can provide assistance and make a first pass at restructuring based on the content itself. Human review is important here, particularly for subjects that require subject matter expertise. But even an initial, imperfect pass can save time.
Migration
When it comes time to actually move the content from A to B, there is almost always a step where the content has to be modified slightly going from one system to the other. Often it’s tweaking HTML — removing invalid markup, or adjusting classes or styles to fit the new design system. Sometimes it’s as drastic as a full conversion from one container format to another. This transformation can be done a couple of ways.
AI-assisted transformation process.
In this process, a developer uses AI to build a transformation pipeline in code which runs against the incoming content. This makes a process that is consistent, repeatable, and idempotent. In other words, if you run the same piece of content through the pipeline three times, you’ll get the exact same output all three times. This is the path of least surprises, but lacks flexibility and requires a bit more work up front.
AI-powered migration.
Here we use an AI agent to do the conversion itself. In this process, the AI agent is the transformation pipeline. It consumes each piece of content, and using instructions that we’ve provided, converts the content itself, making decisions based on the content it’s currently handling. This requires less up front work, but also requires more oversight. Frontier models are astounding these days, but no model is 100% correct every time and even the best models hallucinate occasionally. If you feed the same article in three times, you may get three slightly different outputs. Chances are they’ll all be acceptable, but potentially subtly different in how they are encoded, marked up, or even written. This process requires more human oversight.
Enhancements
Finally, including AI-powered tools in our migration process gives us opportunities to improve and expand on content.
- Generating any missing or incomplete summaries, metadata, or taxonomy. If you’re missing any SEO/GEO data, this is an excellent opportunity to fill in those gaps, and AI can generate much of that with very little intervention.
- Adjusting content for the new realities of the web today. Writing for the web has changed considerably in the last few years. We are no longer just writing for people, but also for AI bots and agents. We can use AI to generate summaries, add questions & answers, and make adjustments to serve all these audiences.
- Auditing content for compliance, completeness, and suitability. As content is analyzed and transformed, we can put AI to further use in generating reports on these different aspects of the content.
Content migration can be as simple as copying and pasting text from one location to another, but more often than not migrations are messy, complex beasts.
However, by using these new tools to our advantage, that messy, complex beast can not only be contained, but also cleaned up. A migration provides an opportunity to derive new value, enhance your content, and prepare it for the new web. That’s where AI helps, shaping a difficult problem into something much more manageable — and a lot more beneficial.
