The framework
The line
There's a line in every business. On one side, a person does the chore when they have time. On the other, something the business owns does it every day.
The line
Two sides
Both sides get real work done. The difference is what's left in the business after the person who did it moves on.
If you're here from the self-check or the audit deliverable, this is the long version of the sentence you already read.
Somewhere in your building there's a report worth real money that runs only when someone has time. Nobody built a job around chasing it — it lands on whoever knows the numbers. That's the before side of the line: the work is real, and it belongs to one person's spare hours.
After the line, the same report belongs to the business. It runs every morning whether anyone is free or not, and it comes out the same way every time. When the person who built the habit moves on, the report doesn't go with them.
The same report, twice
| Before the line | After the line | |
|---|---|---|
| Whose time it needs | One person's spare hours | None — it runs on its own |
| Cost of one more run | Another slice of that person's day | Effectively nothing |
| If that person leaves | The work leaves with them | It keeps running |
| What the business is left with | A faster person | A tool it owns |
The map
Six levels
You've seen what the line does to one report. Here's the whole map: six levels, and the line between them.
- L1
AI Chat
A person uses an AI assistant in a chat window to move their own work along faster. The help is real, and it stops when the window closes.
- L2
Desktop Automation
Small tools speed up one person's computer work, like sorting files or reformatting spreadsheets. The work gets faster, but it still depends on the person who set it up.
The line
Everything before this point makes a person faster. Past it, the business owns something it keeps.
- L3
AI-Built Tools
AI writes the working code once, and the finished tool runs on its own from then on. Anyone on the team can use it, and it comes out the same way every run.
- L4
Connected Workflows
Connected workflows wire tools into the software the business already runs: the ERP, the accounting system, the inbox. Work moves across those systems without a person carrying it between screens.
- L5
Reliable Systems
A reliable system runs a whole job on a schedule and checks its own work as it goes. A person approves anything that matters before it changes.
We renamed this level once already. The old name sounded like a pitch, so it's now the plainest thing that's still true.
- L6
AI-Native Processes
The process itself is rebuilt around what AI makes possible, not a tool bolted onto the old process. It's the rarest level, and only pays where the evidence justifies the redesign.
One fixed point, so the map isn't abstract: the rebate-and-margin tool we built runs inside the Epicor P21 ERP at Kelsan, a multi-state distributor in our own group. Every day it reads the rebate terms and checks each line item against them, on a schedule, with a person signing off before anything changes. On this map, that's a level-five build.
The document
The audit
The six levels show up in two places: this page, and the deliverable an audit hands you.
The audit scores your operation against your own data: which chores sit before the line, what it would take to move one across, and what that would be worth. Level names and numbers stay on this page and in that document. The rest of the site only cares which side of the line the work is on.
The deliverable is a build-or-don't-build verdict and a spec you keep either way — worth having especially if the verdict is don't build. The fee and the timeline sit on the audit page, all on one screen.
FAQ
Questions
What are the Six Levels?
The Six Levels of AI Deployment are six stages of handing work over to AI, from a person using a chat window to a process rebuilt from the ground up around it. The line between levels two and three is the part that matters: it separates making a person faster from owning a tool that keeps working after people leave.
Why does the line matter?
Because it separates two different purchases. Before the line you're buying speed for individual people, and it walks out the door with them. After it you're buying a tool the business owns: it runs every day, and it's still running after those people are gone. The audit exists to price that crossing for one specific chore.
Which side of the line are we on?
Take the self-check on the audit page. It's two minutes, and it tells you which side of the line you're on. Treat it as a self-score: the audit is what verifies the answer against your actual data and puts the verdict in writing.
Is this an AI maturity model?
That's the category, yes. The difference is what it's for. A score tells you where you rank; this map ends in a build-or-don't-build verdict about one specific chore. The full deliverable is spelled out on the audit page.
The first step
Cross the line
You don't need all six levels to start. The audit prices one chore, names what fixing it takes, and hands you a spec you keep.
Fixed fee ($7,500)fixed durationa build-or-don't-build verdict you keep either way