A 5-step self-guided diagnostic. Takes about 20 minutes. Most people find 2–3 misplaced AI tools on the first pass.
Pick the one operational process that consumes the most time — weekly. Draw every step in sequence: who does what, in what order, at what frequency. If you struggle to draw it, that's data.
// aim for 8–12 steps; if you have more than 20, you're mapping too broadly
Go through your process map and put a mark on any step that is: (a) done the same way every time, (b) predictable in its inputs and outputs, (c) not requiring creative judgment. These are your candidates.
// most processes have 3–6 of these; if you find fewer than 2, the process is too high-level
For each marked step, ask: "Could I train someone with clear instructions to handle this and hand me back exactly what I need?" If yes — it's an AI candidate. If no — the process needs more definition first.
// the filter eliminates about 60% of apparent candidates
Of the steps that passed the filter, rank them by how much time they consume per week. The top-ranked item is your first placement target. Start there. Only there.
// resist the urge to place AI in multiple steps at once
Before placing any AI, write down: what does success look like? How will you know if the process got simpler? What metric will you check in 30 days? If you can't answer this, go back to step one.
// no measurement plan = no way to validate = no way to remove if it isn't working
We'll email you a printable version with the five steps, the filter question, and a simple template for drawing your process map.
Great. One placement. Measure for 30 days. Then decide if there's a second. The methodology page explains how to place correctly.
Your process needs more definition first. That's not a failure — that's the diagnostic working. Map one layer deeper and run through the steps again.
That's what the Audit is for. Two sessions where we run through your operation together and build the prescription.