More and more people are handing over their entire computer upkeep — from cleaning out the Downloads folder to sorting notes and files — to AI agents like Claude Code. But there's a curious paradox at play. Folders get tidy enough that "AI runs my computer now" starts to feel true, yet turning all that neatly filed material into a new idea remains just as daunting as ever. This piece traces the problem back to categorization itself, and lays out, step by step, how to get a creative payoff by having your AI agent build connections instead of building categories.
Why Tidier Folders Leave You More Stuck
Most of the tidying AI agents do by default is library-style classification: documents go in the documents folder, receipts go in the receipts folder, everything filed neatly by topic. It looks great, but it has a structural limit — data has no meaning on its own. Even a few thousand perfectly sorted files are still just data; they only become information once you're able to read meaning into them. A category answers "where did I put this?" but never "why does this matter to me right now?" That's why the sense of being stuck survives the cleanup.
Center on Interests, Not Categories
Zettelkasten and the Second Brain — two well-known note-keeping methods — differ in their specific tools but share one core principle: ditch library-style categories, and instead store and link memories around whatever you currently care about. The same clipped article means nothing more once it's dropped into an "economic news" folder, but linked to the interest "the subscription service I'm launching next year," it sits alongside your draft proposal, competitor notes, and yesterday's book excerpt. The first step is shifting your filing criterion from the properties of the material to your own current interests.
Creativity Is the Ability to Spot Invisible Connections
Once material starts accumulating around your interests, what comes next matters most: the ability to find the invisible relationships among that gathered information is, itself, creativity. New ideas don't appear out of nowhere. When two seemingly unrelated notes meet under the same interest, the hidden relationship between them surfaces, and a combination no one else has seen is born. If folder classification is the work of keeping material apart, connection-centered storage is the work of systematically raising the odds of that meeting.
If AI Is Managing Your Computer, Put It to Work on Connections
This is where an AI agent's real value shows up. The biggest barrier to making connections, for a human, is time and memory — and that's exactly what an agent can take off your hands. All it takes is changing how you instruct it. Instead of "put this file in the right folder," try this instead. First, give it your running list of active interests and have it judge which one each new item touches. Second, when it saves something, have it also leave a link to related existing notes or files. Third, have it periodically scan across folders and report pairs of material that look worth connecting. Whether "AI manages my computer" ends up meaning a cleaning service or the groundwork for creativity comes down to this difference in instructions.
A Checklist to Start Today
Here's the order to put this into practice right away.
1. Narrow your current interests and projects down to three to five, and tell your agent what they are. 2. When new material comes in, have it ask "which interest does this touch?" instead of "what kind of file is this?" 3. Every time something gets saved, have it also leave a link or one-line note pointing to related material. 4. Once a week, set aside time for it to report pairs of material worth connecting. 5. Write the reason for each connection yourself, in your own words — because reading meaning into data, turning it into information, is ultimately your job, not the AI's.
Folder cleanup is already an agent's job these days. The question left is one only you can answer: is the AI managing your computer a librarian, or a connector?
