You are already somewhere
on this ladder.

Six levels in three stages. Moving inside a stage sharpens how you work; crossing into the next one changes what you are working on.

Personal

L0–L1

Just you and a chat window. The gains are real, and they stop when you stop.

L0 Getting Familiar

LOOKING FOR USE CASES

“AI sounds interesting, but I don’t know how I’d use it.”

KPIUsage Metrics (Frequency & Trust)

L1 One-Off

Chat: Get advice

“I use ChatGPT to research topics faster. It consistently saves me time.”

KPITime Saved

Team

L2–L3

The way you work with AI stops being only yours. Your team picks it up and runs it too.

L2 Shared Knowledge

Docs: Get personalized consumables

“I stopped pasting context in. It reads our real docs, and what I teach it, my team gets too.”

KPIQuality Metrics (Error Rates, rework)

L3 Shared Skills

Work: AI completes tasks for you to modify

“I have AI doing work for me. The tedious stuff is taken care of.”

KPIWorkflow Metrics (Cycle time, Throughput)

Agents

L4–L5

Work runs without being asked for. Your job becomes directing it and checking it.

L4 Autonomous Agents

Auto: AI systems get work done for you

“My job is to build the system that makes the output, not the output.”

KPIFinancial Metrics (Cost, Revenue)

L5 Cross-Team Compounding

Grow: AI builds systems driving outcomes across teams

“Across teams, things just run. Gaps get flagged before I see them.”

KPI24/7 agent uptime

A starting point you can name

If you want to get somewhere, you first have to know where you are standing. A level gives that a name, and the name comes with a short list of what is actually worth your attention at that stage.

A snapshot worth arguing with

It reflects how you worked recently, which is usually enough to surface a gap or a strength you had not put words to. Either one is a conversation starter about where to go next, whether that conversation is with your team or just with yourself.

Permission to ignore things

Most of what gets written about AI is aimed at somebody else's level. Yours tells you what is relevant to your journey today and what you can safely leave until later. The training list works the same way: 34 picks in the library, filtered down to the ones that fit where you are.

A next action instead of a reading list

Every level ends in one concrete move with a time block on it, from ten minutes to an hour. Leading with work you can actually apply is how you stay out of the sea of hype.

It does not stop at prompting

The upper rungs are specified to the same detail as the lower ones. L4 asks for a system two other functions can feel and one workflow whose output you have stopped reading. L5 asks whether it all held while you were gone for five days. If you are already running agents, those are the lines to measure against.

• My AI LevelTen situational questions about what you actually did last week, not how confident you feel. It returns a level, the evidence it read to get there, and one next move sized to fit.
• AI Level GuideThe guide itself. Five views of the same six levels, so you can come in by whichever door makes sense.

  • 01Company 1-PagerWhere the organisation sits on the curve, plus the printable one-pager: six level cards on a single sheet, sized to go straight onto a slide.
  • 02The Ladder at ScaleThe same six rungs read against one workflow, one person, one team or the whole company. Scope is fractal here: the pattern repeats whatever you point it at, and the level you are depends on what you are measuring.
  • 03Full Success PathEvery level written out in full, every field, side by side. This is the long version the rest of the guide is drawn from.
  • 04Examples by TeamWhat each level looks like in your function rather than in the abstract.
  • 05ImpactYour level set against what it actually produced, because climbing the ladder and delivering something are two different achievements.
• TrainingLearning and practice picks filtered to a single level, each level ending in one thing to go build. Set the filters, copy the address, and whoever opens it sees exactly the set you meant and nothing else.
• Cheat SheetsFour sheets, front to back: what these tools actually do and which one fits the job, what never goes into them, how much thinking room to give the model, and the version-control vocabulary your team uses without explaining.
• Facilitator GuideFor when you are the one helping other people move: a session format, and the questions worth asking a room sitting at each level. Opens once you have taken the assessment.
• Quest BoardPost a piece of work, claim someone else's, publish what you built. Still a draft: boards live on your own device and travel by link rather than sitting in a shared queue.

1. 01

   Pick your honest baseline

   The level whose description feels earned rather than aspirational is your real one. The flattering answer is the only wrong answer here.

2. 02

   Aim one level up

   Not the top. The next one. Read its graduation gate and treat that as the target, since it is the only line that tells you when you have actually crossed.

3. 03

   Pick one change you can repeat

   Something you can do again next week without deciding to. That is what moves a level; a heroic afternoon does not.

• How long does the assessment take?

  Ten questions, around five minutes. Most are pick one from a list. Two of them ask you to type a sentence, and both can be skipped.

• Do I need to be using AI already?

  No. L0 is a level, not a failure to have one, and its first action is ten minutes long. Getting familiar is the work at that level.

• Does it matter which AI tool I use?

  No. The levels describe how you work, not what you hold a licence for. The question about which tools you use carries no weight in the score at all; it is there so the guide knows what to point at.

• What if the level I get feels wrong?

  Read the evidence the report shows and decide for yourself. Your level is capped at the highest gate you actually pass and ties resolve downward on purpose, so a level that feels low often traces to one unticked gate rather than to the whole read missing. Take it again whenever you want.

• Who sees my answers?

  There is no sign in and no account. Your result is saved in this browser, so another device or a cleared browser starts fresh. Responses are recorded against a random per-device id rather than your name, and the name and email fields only appear if you choose to make a shareable link.

• Is this a ranking?

  No. There is no leaderboard, no manager view, and no surface anywhere in the guide that rolls results up or sets one person against another.

• Can I send this to someone else?

  Yes. A share link opens your saved report read-only for whoever you send it to, and the company one-pager prints to a single sheet sized to go straight onto a slide.

• Where did the six levels come from?

  Conversations about how AI transformation actually lands at scale, at the companies named at the top of this page, and the rubric those conversations turned into. It is a working model, not a standard.