Your agency's client folders are the most valuable part of its AI

In the last post I explained the AI problem most agencies haven't hit yet: once AI is writing for the whole client list, everything piles into one memory, and it starts grabbing the wrong client's details. The fix comes from a very old idea: a well-kept filing cabinet.

Here's the payoff first. When one client bans a word, you tell the AI once, and no other client's posts are touched. When Google changes a structured-data rule, you update it in one place, and every client's next schema follows it. The rest of this post is how that works.

What does the research say?

In March 2026, Jake Van Clief and David McDermott published Interpretable Context Methodology: Folder Structure as Agentic Architecture.¹ Their central idea is that the way you organize folders can be the thing that directs an AI through a job.

Each folder holds one stage or one area of work, plus a short plain-English file that says what's in the folder, what the AI should read, and what it should produce. The AI opens only the folders the current task needs. In the paper's examples, each step reads a few thousand words of focused material instead of tens of thousands of words of everything.¹

Interpretable means a person can read all of it. Every instruction and every draft along the way is a file someone at the agency can open, check and correct, not something locked where only a vendor can see it.

How should an SEO agency organize its files for AI?

Here's a simplified layout for an agency that does content, on-page, schema and paid search. Each INSTRUCTIONS file is a page or two of plain English that an SEO manager could have written.

agency/
  INSTRUCTIONS.md              how this agency works, who approves what
  services/
    content/
      blog-post/
        INSTRUCTIONS.md        brief → draft → QA checklist
    on-page/
      meta-titles/
        INSTRUCTIONS.md        length limits, format, what never changes
    schema/
      INSTRUCTIONS.md          which markup for which page type
    paid/
      rsa-copy/
        INSTRUCTIONS.md        headline and description limits, review steps
  clients/
    example-dental/
      INSTRUCTIONS.md          voice, banned words, service area,
                                approved keywords, approval contact
      pages/                   current page copy, drafts, schema
      monthly-reports/
    example-roofing/
      INSTRUCTIONS.md
      ...

When a writer asks for this month's posts for Example Dental, the AI is given the agency's instructions, the blog-post procedure and that one client's folder. It isn't given Example Roofing's service area or another client's keyword list, so it has nothing to borrow from. Making sure of that is part of the setup, not something folder names do on their own.

The same layout works for web design, local listings and reporting, each with its own procedures.

What happens when a client's instructions change?

This is where the setup pays off day to day. Say an account manager tells the AI, "Example Dental doesn't want the word 'painless' anywhere, and they've stopped serving the north county office." Because each client's rules have their own home, that note goes into Example Dental's file and nowhere else. Every Example Dental draft after that follows it. Every other client is untouched.

Or: Google changes which structured data it uses for a rich result. That change belongs in the schema folder, once. Every client's next schema pass follows the new rule, and no one has to remember to tell each writer.

The same works for a single page ("never change the H1 on the homepage"), a service, or the whole agency. One sentence, stored once, used every time it applies.

To be precise about the source: the paper covers step-by-step work with a person reviewing each stage. Routing everyday corrections into the right file is how I apply the same principle to an agency's ongoing work.

Who owns the files, the agency or the vendor?

The AI program that reads these folders can be replaced, and so can the model behind it. The folders, instructions and procedures are the part that belongs to your agency. They're plain text files in your own storage, such as the agency's Google Drive, that you can back up, copy and move.

That matters more than it sounds. When an AI setup lives on one person's laptop or inside one person's chat history, a wiped computer or a staff change takes it with them, and the next person starts from zero. When it lives in the agency's shared files, it stays when people come and go. A new AI tool still has to be connected and checked, but the agency's know-how doesn't start over.

Is this hard for staff to use?

Simple to use. Careful to set up.

Day to day, staff don't learn anything technical. They ask for what they need in plain English. The careful part is the setup: mapping how the agency actually delivers, deciding where each kind of knowledge lives, and writing the first set of instructions for each service and each client. That has to be right at the start, because a messy setup gets messier with every client you add.

That setup is the work I do for agencies. In the next post I show what it looks like across an SEO team's week.