The AI Proficiency Scale sorts people into three levels: Beginner (uses AI rarely, or with uneven results), AI Champion (gets repeatable results that others use) and Systems Builder (redesigns the work itself, and knows when not to use AI). Each level comes with one question that moves a person up. The levels tell you where someone should start, not how good they are.
It's the scale we use in our training and our talks, and it's about as much framework as we think most leaders need. We'll be using it with a room of project managers this week. Here's the whole thing, how to place people on it, and why it matters for anyone leading a team through AI.
What are the three levels?
Beginner
Uses AI rarely, or with uneven results. A Beginner might have tried ChatGPT a few times, gotten something useful once and something wrong the next time, and drifted back to the old way. Or they use it daily for one narrow thing and nothing else.
The question that moves a Beginner up: "What could help me today?"
AI Champion
Gets repeatable results that others use. A Champion has found a few things AI does reliably well in their job, and their results show up in other people's work: a template the team relies on, a prompt a colleague borrowed, a summary format everyone now expects.
The question that moves a Champion up: "What repeats, and who else could use it?"
Systems Builder
Redesigns the work itself, and knows when not to use AI. A Systems Builder looks at a process and asks why each step exists before deciding what to automate. They're as likely to remove a step as to add a tool, and they can tell you where AI doesn't belong.
The question that moves a Systems Builder further: "What system should exist?"
The AI Proficiency Scale at a glance
| Level | What it looks like | The question that moves you up | One next step |
|---|---|---|---|
| Beginner | Uses AI rarely, or with uneven results | What could help me today? | Ask AI to teach you one thing from your work, fifteen minutes a day. |
| AI Champion | Gets repeatable results that others use | What repeats, and who else could use it? | Turn one win into something a teammate uses without you. Before you automate a step, ask why it exists. |
| Systems Builder | Redesigns the work itself, and knows when not to use AI | What system should exist? | Take one team process apart with the people who do it, and bring the before and after to a leader. |
Is the scale a grade?
No. Levels describe where to start, not how good someone is. A senior partner can be a Beginner. A first-year analyst can be a Systems Builder. Neither fact says anything about their judgment, their expertise or their value.
Levels are also specific to an area of work. You can be a Systems Builder in your reporting and a Beginner in client communication. When you place someone, place them for the work you're trying to improve.
How do you place a person on the scale?
Self-ratings are unreliable. In our experience, some of the strongest users underrate themselves, and some people who've tried AI twice rate themselves highly. So we ask for evidence instead. Four questions do most of the work:
- "The last time you used AI for work, what was the task, what did it produce, and what did you do with the result?" This separates occasional use from real use.
- "Has anyone used something you made with AI without you there?" This is the line between a Beginner and a Champion.
- "Name one process that changed because of AI. Not faster: changed." This is the line between a Champion and a Systems Builder.
- "Name something you decided not to use AI for, and why." Systems Builders answer this one quickly and specifically.
Then one rule: when the evidence sits between two levels, choose the lower one. Starting too low costs you a week. Starting too high costs you the whole effort, because the person hits a wall, decides AI isn't for them, and stops.
What does the scale look like for the whole organization?
Individual levels add up to an organizational stage. We use four:
- Scattered. Some people use AI, alone and invisibly. A few people hold everything together.
- Supported. Approved tools and guidance exist, and some people are getting good results, but the wins stay personal.
- Shared. Wins travel between people and teams, and ideas reach a decision in days.
- Systematic. Redesigned processes don't depend on any one person.
If you want to place your organization, our AI readiness assessment walks through seven questions that map to these stages.
Why should leaders use a proficiency scale?
A proficiency scale turns AI adoption from a guess into a plan. Most leaders know roughly how many AI licenses they've bought and very little about how their people actually use them. A three-level scale fixes that cheaply. You take a census, not a guess: ask everyone the four evidence questions, place them at a level for the work that matters, and you'll know how many Beginners, Champions and Systems Builders you have. That census tells you what training to buy, who can coach whom, and which processes are ready to redesign. It also gives everyone a shared language, so "we need to get better at AI" becomes "most of finance is at Beginner, two people are Champions, and nobody is redesigning the close." And because the levels describe a starting point rather than a grade, people will tell you the truth. That's what makes the census accurate.
Three more reasons it earns its keep:
One-size training misses most of the room. A single session for everyone usually lands only with the people near the trainer's level. Beginners get lost, Champions get bored. This is a big part of why so many businesses end up with AI licenses nobody uses.
AI proficiency isn't process expertise. The best opportunity-spotter in your company may be a Beginner. They know where the work hurts. Pair them with a Champion and you often get a better project than either would find alone.
It shows you who to invest in. Champions are your multipliers. Systems Builders are scarce, and they need time and a leader's backing more than they need another tool.
How is this different from other AI proficiency frameworks?
There are good published frameworks, and they're built for different jobs. Trust Insights' six levels of AI proficiency map levels to career tiers, from individual contributors through managers to leaders, and its authors make the same point we do: a higher level isn't inherently better. Larridin's AI proficiency framework uses five levels, from "Search Replacer" to "AI-Native Orchestrator," scored across nine dimensions from observed behavior, which suits large organizations that want measurement at scale.
Ours is deliberately smaller. Three levels, four questions and one rule are something a manager can use in a hallway conversation, without software or a survey. For a business of 30 or 80 people, that's usually the right size.
Where to start
Pick one team. Ask each person the four evidence questions, place them honestly (lower when in doubt), and give each person the one next step for their level. In a month, ask again.
If you'd like help running the census or training each level, we teach this scale in our onsite and virtual AI training for teams across Iowa, and use it when we work with leaders through Advise, Educate & Lead. For individuals who want to start with their own habits, our free AI Enablement exercise is a good first fifteen minutes. And when you're ready to bring it to your whole staff, our AI training for teams and leaders is built around these three levels.