We built this internally, grounded in research, because we needed a calm way to find our starting point on AI. It's not about where any leader "is" or how they score. It's about us finding the starting point together, then iterating from there.
Calibrate, move, recalibrate.
Start here →Two things worth carrying into the rest of this. They shape what it's really for.
This isn't about understanding AI in the abstract. It's about how we use it, how visible we are in showing the things we're doing, big and small. Curiosity in practice. People watch what we do as leaders, not so much what we say.
Important not to mandate but to steer. Give our teams a direction: here's the tool we use, here's what to use it for, here's how to get better at it. Then from there we look for the people already exploring, and back them. They're how it spreads.
Pick one. Or move through all four over a few weeks.
Run the slider on your team as you see them. Note the gaps.
Send each person the link. They rate on their own device. State saves locally. Compare the reads when you next meet.
Take your result into your next 1:1. Ask for what you need to move the team.
Scores shift when conditions change. See where the dot has moved.
Scroll, or use the rail on the right to jump. Eleven sections, ~20 minutes if you read straight through.
Toggle bottom-right corner. Scales sliders and scenes up for projecting in a room with your team.
Auto-saves to your browser as you go. Three ways to share when ready: image, link, text. Each person's state stays private.
Worth re-running quarterly. The value isn't the score, it's where the dot moves over time.
It's not who people are. It's what's around them.
BetterUp Labs ran the regression. Personality traits barely move the needle. The four things that dominate are environmental: how leadership communicates about AI, whether it's mandated, how much psychological safety exists, and how much agency people feel over their work.
All four are conditions. All four are shapeable.
A diagnostic, not a verdict. Walk all four honestly, then run the assessment below.
Tap to expand. Read all four honestly.
Using AI deliberately. Coaching still happens. People are still doing the thinking; AI helps them think better.
Real examples shared in team meetings. Stuck problems getting new angles. Nobody afraid to say "I used AI for this."
Protect what we've built. Share our examples upward so others see what's working.
Strong human side. Coaching happens. But AI adoption is thin. Output stays flat.
Good 1:1s. But people doing work AI could improve, and not reaching for it. Permission feels ambiguous.
Make AI use visible. Show our prompts, not just our output. The foundation is there; the tools need permission.
Using AI heavily, human side thinning. Volume up, workslop up. Work reads as if nobody cared.
Generic-sounding Slacks. Polished decks saying nothing. Lower energy in the room.
Coach through AI use, not around it. Make shipping slop unacceptable.
On autopilot both ways. Risk isn't visible until someone else moves.
Flat meetings. Output that doesn't embarrass but doesn't surprise. No visible growth.
One visible commitment on each axis. Pair with one teammate. Movement creates movement.
Concrete moments, not abstract categories. Tick the ones that happen on your team today. Section 9 goes deep on the two hardest.
Nobody treats it as a confession.
Not a demand. An example. Visible on purpose.
Polish doesn't equal correctness.
Same reaction as any mistake: what did we learn?
Blocked on the calendar. Not "if you have spare capacity."
In real work, not just the onboarding deck. They pick it up by osmosis.
Before the tool is picked, not after.
Nobody marvels at it. It's just how work happens.
Speed questions paired with limits and risks. Out loud, in the same conversation.
A teammate you ask when you're stuck on how to use a tool. Help is named, not hidden.
Eight questions. The dot moves as you slide.
Four questions on AI investment. Four on investment in people.
An uncomfortable result is still a useful result. You wanted to know where you actually are.
Nothing leaves your browser unless you choose to share. Save your read as an image to drop into Slack or email. Or copy a share link your lead can open to see the same dot on the same quadrant.
A real result. So the framework doesn't stay abstract.
I ran this on my own team: 53 on AI, 80 on people. Just inside Calibrator, but the shape tells the real story: strong on people, still early on AI.
Two months ago it would have been quieter. Now the team's at a sharing stage: small groups trying things, asking each other for feedback on prompts. The motion is real. Team-wide rhythm is further out.
The scenes showed me where to start. Most of the people-side ones were ticking. The gaps were protected time (scene 5) and the augment-or-automate habit (scene 7). That's what we're working on next.
The score isn't the point. The conversation is.
Of the ten scenes, two tend to be the hardest. Also the two with the biggest payoff.
People can say "I don't know how this works" without social cost. Harder than it sounds, because we're used to teams where confidence reads as competence. With AI, that confidence reflex gets people stuck.
Share our own confusion first. "I tried this prompt and got garbage. Here's what I changed." Make the effort visible, not just the output.
Open the next 1:1 with "where have you been stuck with AI this week?" and listen. Don't coach. Normalise the stuck.
An hour on the calendar that isn't consumed by other work, defended when priorities arrive. Every team says they do this. Very few do.
Take the time first ourselves. Block it, don't move it. If we're always too busy to experiment, we've answered why the team isn't either.
Block 60 minutes labelled "AI practice, do not schedule over." Defend it for four weeks. Tell the team why.
A short read on why we're starting here, and what it sets up.
If AI is going to be part of how every team works at Frontify, we need a shared read of where we actually are first. That's what this is for. Not a verdict. A baseline.
It's not really about the score. It's about us all using the same words for the same things. Once 60+ leaders have run this, conversations about AI get easier: where to invest, what to build for managers, how we hire, how we talk about it across the company.
Where we're trying to get to: 400 people doing the work of closer to 600. Every one of us doing things we couldn't a year ago. Not by replacing people with AI. By giving people the tools and the conditions to be properly good at their work. That's why we're starting in People before asking anyone else to.
Credit where it's due. Go deeper if you want to.
The research below shaped my thinking. Listing it so you can go to source and judge it for yourself.
Where this took shape. Includes the full Case for Conditions PDF, session recordings, and supporting briefings.
betterup.com/agents-of-changeOxford (De Neve), BetterUp (Niederhoffer), Stanford (Hancock). Maps the augmentation vs automation divergence across six adoption phases. The four-posture framework (Calibrator, Automator, Traditionalist, Disengaged) and the conditions regression both come from this work.
betterup.com/blog/ai-investment-roi-human-capitalTangible tech investment is roughly one-ninth of what AI value actually takes. The other eight parts are organisational rewiring, process redesign, skill-building.
Here's what your read looks like right now. Your answers stay saved on this device, so you can come back to it. Re-run it in a quarter to see what's moved.