Hand AI the Whole Folder

There’s a moment every heavy AI user hits: the project that’s too big for the chat window. Not too hard. Too big. Dozens of files, months of work product, logic spread across spreadsheets that only make sense in sequence. Uploading them one at a time to a conversation is a nonstarter.

One of my members hit exactly this. Years ago he completed a long, complex client engagement that produced a thick folder of spreadsheets, each file one step in a process that lived mostly in his head. His question was better than he realized: could AI go through the whole folder, figure out the sequence from the file dates and names, and reconstruct the logic I used?

Yes. But not in a chat window. Here’s the workflow.

Move the work to your desktop

The AI tools you use in a browser also come as desktop apps, and the desktop version changes the game for file-heavy work: instead of uploading documents into a conversation, you point it at a folder on your computer and it works with everything inside.

One rule before anything else, and I treat this as non-negotiable: never give a desktop AI access to your whole computer. Create one dedicated folder for the project. Grant access to that folder and nothing else. Everything it needs goes in, everything it produces stays in, and the rest of your machine stays out of reach.

Brief it like a new hire

Drop your files in the folder, then write the opening prompt the way you’d brief a smart new employee on day one. Tell it what the files are: in this folder are fifty spreadsheets from a client project. Tell it the naming logic: files starting with X are the before state, files with Y are the after. Tell it what you want back: reconstruct the process, map the before to the after, document the logic as a repeatable method.

Then end the brief with the sentence that separates good outcomes from frustrating ones: before you start, do you have any questions for me?

Complex projects always have ambiguity. Asked up front, the AI surfaces the ambiguity as questions. Not asked, it guesses, and you find the wrong guesses three hours later.

The payoff: your old work is a product

Here’s why this workflow matters beyond cleanup. Buried in your past engagements is your methodology: the sequence, the decision rules, the questions you ask, the way you untangle the mess. You’ve never written it down because you didn’t have to. It’s in your hands.

AI reading the whole folder can extract it. The output isn’t just documentation. It’s the raw material for a sellable asset: a repeatable framework, a discovery checklist, a paid diagnostic, a productized version of the thing you currently reinvent for every client.

And when the AI inevitably tells you your productization idea is great, don’t take the compliment. Ask it what could go wrong. Ask why this might be a bad idea. AI agrees by default, and an idea that survives the criticism round is worth a lot more than one that was merely applauded.

The chat window is for conversations. The folder is for careers’ worth of work. Once you learn the second mode exists, a lot of impossible projects turn into weekend ones.