Free resource · No email required

Where AI belongs in your business.

A five-minute triage before you build anything. Most businesses lose money on AI the same way: an elaborate AI tool for a job a spreadsheet, a simple rule, or a person already does better and cheaper. This page helps you tell the difference.

The shape of a system that actually works

In the systems I have built and audited, a healthy automated process is mostly ordinary, boring, reliable parts:

60% PLAIN SOFTWARE + DATA
30% SIMPLE RULES
10% AI

The rule of thumb from real builds - most people flip it upside down and reach for AI first

The 60 percent is spreadsheets, databases, calculations, the tools you already use. This part does not guess: a formula gives the same answer every time. The 30 percent is "when this happens, do that" - automations moving information from one place to another. The 10 percent is the thin AI layer on top that handles judgment, reads messy information, and writes in a human voice. Reaching for AI first is where the cost and the headaches come from.

The three questions

For any task you are tempted to "put AI on," ask these in order. Stop at the first yes.

  1. Is it the same every time? A fixed calculation or lookup belongs in a spreadsheet or simple software, not an AI. A formula does not make things up. An AI sometimes does.
  2. Is it a clear if-this-then-that rule? If you can write it as "when X happens, do Y," it belongs in a simple automation. Cheaper, faster, and it never has an off day.
  3. Does it need judgment across messy, unstructured information? Reading a pile of emails and summarizing what matters. Drafting a reply in your voice. Spotting the odd one out in a stack of documents. This is where AI earns its place. If you got here, AI is the right tool.

The cost check

Before you build, price the AI against a person. A capable person handles complex, fuzzy work cheaply. If the monthly cost of an AI tool is more than it would cost to have a person do the same task, the person is the better answer. "We can automate this" is not the same as "we should."

Your do-not-build-yet list

This is the most valuable part, and the one everyone skips. Write down the tasks that should stay manual for now, or stay with a tool you already pay for. Naming what NOT to build yet usually saves more money than anything you do build. And it is a "yet" on purpose: this list is about your starting point, not a verdict on the technology. What made no sense to automate this year is often the natural next step once the foundation under it is solid.

Triage your own tasks

List the tasks you were thinking of using AI for. Run each through the three questions and the cost check, then decide. (Print the page and fill it in - that's what it's for.)

TaskSame every time?A simple rule?Needs judgment?Decision (fills itself in)

Decision options: spreadsheet · automation · AI · keep manual · keep current tool

Type right in the table - answer the three questions with yes or no and the decision column fills itself in. Notice what just decided it: a simple rule, not AI. That is the whole point of this page. If you answered yes to an early question AND yes to judgment, the rule will tell you to split the task - most real work is a routine core with a judgment slice, and they route differently. That split is half of what an analysis does. One honest caveat the rule cannot see: when a task lands on AI, the cost check above still gets the final vote - and for the high-stakes ones (the email to your biggest client, the call on what to promise), AI drafts and you decide. The judgment stays yours. Your entries save in this browser only - nothing is sent to us. Print the page and they come with it.

AI's reach is not fixed. It grows with the person operating it.

The bottom line: you are not trying to become an "AI business" that bolts AI onto everything. You are trying to build a good business that uses AI in the few places it is genuinely better right now. What AI does in your business a year from now will be bigger than what this page maps today - that is exactly why the order matters. The companies that get the most out of AI are honest about their starting point, build the boring, reliable layer first, and let the AI layer grow on top of it.

If you want a second set of eyes on your list, that's exactly what the analysis is.