Many Branches

Receipts · Article 2 of 2

Could a stranger run your firm?

The real AI-readiness test has nothing to do with picking tools. If a competent stranger could not operate your firm tomorrow morning from what is written down, no assistant can either.

1 · The question under every AI question

The question we are asked most often is some version of which AI tools should we buy. It is a reasonable question, and it is almost never the useful one, because the useful one sits underneath it and rarely gets asked at all.

What would your firm look like if a competent stranger had to operate it tomorrow morning, from your documents alone?

Sit with that before reading on, because your first honest answer is the most valuable diagnostic in this entire library.

In most firms we meet, the stranger would stall before lunch. The client history lives in three people’s memories. The way work actually flows, as distinct from the way the old process document says it flows, lives in habits nobody has ever described out loud. The pricing logic, the quality bar, the exceptions that matter: heads, all of it. The firm works because the people who hold it in their heads come in every day.

That is not a criticism. It is the default condition of every organization that has never been forced to explain itself, and until recently there was no forcing event.

2 · The stranger has arrived

There is one now, and it is the thing your vendors are selling you.

An AI assistant is precisely that competent stranger. It reads everything you give it, works at any hour, and knows nothing about your firm beyond what is written down. Every firm that pilots one discovers its true documentation state inside the first week, because the assistant’s answers are exactly as good as the written record it was handed and no better. The tool did not create the gap. It made the gap visible, priced, and suddenly urgent.

Which is why we distinguish between a firm that is AI-enabled and one that is AI-ready.

AI-enabled means somebody bought subscriptions. The chat window is open, the operating knowledge is still in heads, and the assistant produces generic answers wearing your letterhead. AI-ready means the firm’s knowledge, its rules, its history, its way of doing things, exists in a form a reader with no context can act on. The first is a purchase. The second is a property of the firm itself, and it is the property every tool you will ever buy depends on. Buying tools first and legibility later is how firms end up with a drawer of abandoned pilots and a settled conviction that AI is overrated.

3 · You have met this test before

If you have read the rest of this library, you have already met the stranger test five times in five costumes, and noticing that is what this closing pair is for.

The tool inventory that opens our governance series is the stranger test applied to your software: could somebody who is not you say what is in use, on whose accounts, touching which work? The written, tiered policy is the same test applied to your rules: could a new contractor learn what is allowed without asking three people and getting three answers? Our roadmap article states it for your technology plan almost word for word, since a new leader should be able to pick the document up and run the firm’s technology from it. Stage independence is the test applied to your vendors: could anybody else execute this plan, or is it a leash? And the map of where facts live, from our article on federated search, is the test applied to your knowledge itself: which system is the truth for commitments, for work state, for the current version of anything?

One question, five surfaces. Governance, it turns out, was never really about AI. It is about whether your firm can explain itself, and AI is simply the first reader demanding that it do so in writing.

4 · We are the experiment

We hold this opinion with some confidence, because our own firm is the ongoing test of it.

Many Branches is two people, and our daily working colleague is an AI assistant that remembers nothing from one session to the next. Every morning it arrives knowing only what we have written down: our procedures, our standards, our map of where the firm’s facts live. The firm runs anyway, every day, because we pay the legibility cost continuously instead of hoping never to pay it at all. When we say the stranger test is passable, we are not speculating. We take it every morning, and our documentation gets graded by the least forgiving reader there is.

One asymmetry belongs in the open, though. We had it easy. We compiled our firm from a blank page, writing things down as we built them. An established firm arrives carrying decades of tacit knowledge nobody ever wrote down, held by people with jobs to do, and document everything is useless advice at that scale.

Which is exactly why our engagements begin with interviews and an inventory rather than with tools. The first weeks are the stranger test administered gently, by people who have run it before, and the first deliverable is the legibility itself, in a form your firm keeps whatever happens next.

5 · The Monday morning version

You do not need us to start.

Pick one process your firm depends on and have the person who knows it best write it down so that a stranger could run it. Then hand it to a capable colleague who was not in the room, and watch where they stall. Each stall marks a place the process actually lives in somebody’s head, and the stall list, an hour’s work, is a truer picture of your AI readiness than any vendor questionnaire will ever produce.

Then notice the property that makes this work worth doing regardless of what happens with AI. Everything you write down for the stranger also onboards your next hire, survives your key person’s vacation, raises the price an acquirer will pay, and shortens every audit and questionnaire you will ever face. Legibility compounds, and it is the rare investment that pays out whether or not any particular technology delivers on its promises.

Tools will keep changing under all of us. A firm that can explain itself will be ready for each one. A firm that cannot will be sold another pilot.

That closes the receipts, and with them this library. The front door, as ever: a short conversation, a fixed-fee assessment, and findings you own either way.

Start a conversation

The stranger test opens our fixed-fee readiness assessment: we map what is written down, what lives only in heads, and what closing that gap would earn you, before any tool gets discussed.

If your firm already passes the stranger test, AI adoption will feel to you like a deployment rather than a project, and you should hold every vendor you meet to that same standard.

Start a conversation