Content Strategy: AI as a Pattern Detector

There is a version of AI-assisted content work that gets most of the attention: faster drafts, automated outlines, and generated social copy. Useful, yes. But it’s also the most surface-level application of a technology that is genuinely powerful when you point it at a harder problem.

That harder problem is analysis. And it’s where AI has quietly become one of the most valuable tools in my own workflow. In fact, when I teach my AI-Assisted Writing course, I share examples of where AI surprised me in its sometimes deeply nuanced, sometimes flagrantly incorrect analysis. But it’s proven, overall, to be more useful than error-prone.

Case-in-point: I’ve been working across multiple book drafts simultaneously for the past three years. Moving between them, tracking how ideas have evolved, identifying where arguments overlap or contradict each other, spotting the gaps that only become obvious when you step back from the writing itself — that kind of editorial oversight is genuinely difficult to maintain at scale. Feeding those drafts into AI changed how I work. Not because it writes for me, but because it reads across everything I’ve written and surfaces patterns I can’t see from inside the work. It functions less like a ghostwriter and more like a sharp developmental editor who has read every draft, remembered all of it, and has no emotional attachment to any of it.

In this latest article in my ongoing Content Strategy series, I want to make the case for that different application: using AI not primarily to generate content, but to analyze it.

Proximity Is the Enemy of Perspective

When you’ve been producing content for any significant length of time, you develop blind spots. Topics you’ve covered so many times that you no longer notice the repetition. Angles you’ve never taken because you unconsciously assumed you had. Formats that stopped performing months ago but still feel like a safe default.

These patterns are real and they have strategic consequences. Redundant content dilutes your authority on a topic rather than reinforcing it. Uncovered gaps leave audience questions unanswered and search opportunities on the table. Overused narratives train your audience to skim because they’ve seen the shape of your argument before.

The scale of this problem becomes vivid when you’re working on something as complex as a book. Sharon Weaver and I are currently writing Volume 1 of the Project Failure Files, drawing from nearly three years of notes, more than two years of podcast episodes, and ideas captured across dozens of OneNote entries with no thought given to chapters or organization at the time. We started with an unwieldy list of topics and themes, and have been finding and refining the patterns ever since. When you’re ninety episodes deep and trying to pull that material into a coherent chapter structure, the organizational challenge alone becomes formidable — and the risk of missing something important, or inadvertently covering the same ground twice, is very real.

The problem isn’t discipline or effort. It’s proximity. When you’re the one who created the work, you’re too close to it to see it clearly. That’s not a flaw in the process. It’s just how human cognition works. And it’s exactly where AI earns its place.

What AI Is Actually Good At Here

AI tools are well-suited to the kind of analysis that humans find tedious and error-prone at scale. Feed a language model a meaningful sample of your existing content and ask it to do analytical work rather than generative work, and the outputs become genuinely useful.

A few specific applications worth building into your content workflow:

• Theme clustering. Ask AI to group your existing posts, articles, or newsletters by underlying topic or argument. The clusters it identifies often reveal both areas of genuine depth and areas where coverage is thinner than you assumed.
• Redundancy detection. Ask it to flag pieces that cover substantially similar ground. This is particularly useful if you’ve been publishing for several years and your archive has grown large enough that you can no longer hold it all in your head.
• Gap identification. Once you have a clear map of what you’ve covered, ask AI to identify the adjacent questions your content doesn’t answer. Cross-reference those gaps against what your audience is actually asking, through comments, replies, search queries, and direct conversations.
• Narrative pattern analysis. Ask it to identify recurring structures, arguments, or framings across your content. If you’re opening every piece with a problem statement and resolving it with a three-part framework, that pattern is worth knowing about, both because it might be becoming predictable and because it might be a genuine strength worth leaning into more deliberately.
• Performance trend synthesis. Feed AI your analytics summaries alongside your content inventory and ask it to identify correlations. Which topics consistently outperform? Which formats have declined over time? What does the data suggest about where your audience is most engaged versus where they’re dropping off?

The Human Judgment Layer

None of this replaces the editorial decisions that require context AI doesn’t have. It doesn’t know which gaps are worth filling based on your business priorities. It can’t tell you whether a redundant piece should be consolidated, updated, or retired based on how it fits your broader narrative. It won’t understand why a particular underperforming piece matters strategically, even if the numbers don’t show it yet.

What it can do is surface the raw material for those decisions faster and more completely than you could on your own. Think of it as doing the reconnaissance so you can make better calls about where to deploy your actual creative energy.

The workflow that works best is sequential: run the analysis first, review the outputs with your own judgment applied, and then decide what to create. Not the other way around. Starting with AI-generated content and then trying to figure out where it fits is putting the cart before the horse. Starting with AI-generated insight and then writing deliberately toward the gaps it reveals is a genuinely stronger approach.

A Different Question to Ask

The next time you open an AI tool to work on content, try asking it something analytical before asking it to write anything.

What themes dominate my recent content? Where am I repeating myself? What questions is my audience asking that I haven’t answered? What does my publishing pattern suggest about where my strategic focus actually is versus where I think it is?

The AI tools will not do the work for you. You still need to use your judgment and experience to validate the patterns and then decide what to do with that information — expand on an idea, edit a section, add a chapter, retire a narrative that has run its course. Think of AI as the tool that surfaces the right questions. The answers, and what you do with them, are still yours to own.

That discipline is what makes whatever you write next more useful, more focused, and more worth your audience’s time. And it’s a better use of the technology than faster first drafts.