How much of your job
can AI actually do?
Not the headline number. The real one.
Map your daily tasks. See which ones are in the automation zone —
and which ones are protected by the informal knowledge, relationships,
and human judgment that AI agents can't access.
9,018
Assessments completed
29%
Average automation exposure
The Framework
Most analyses get it wrong.
Here's why.
They classify jobs as "automatable" based on whether the input and output are digital. But that's only the surface layer. We assess three:
Formal I/O
Is the task digital-in, digital-out? This is the only layer most analysts look at. It defines the theoretical automation zone.
Contextual Substrate
Does the task actually depend on hallway conversations, gut feelings, 15 years of pattern recognition? A task can look digital on paper but depend entirely on informal knowledge. AI has zero access to this layer.
Social Function
Does performing this task also build trust, maintain relationships, or signal presence? The lunch with a client isn't just information transfer — it's the relationship itself. Remove the human and you lose the value.
An AI agent can only replace a task when all three layers are either digital or irrelevant. That's a much smaller surface than the headlines suggest.
How It Works
A conversation, not a survey.
01
Tell us your job
Enter your job title. Our AI proposes typical tasks for your role. You confirm, edit, or add what's missing.
02
Watch your profile build
As you interact, a visual map of your job assembles in real time. Each task is classified across all three layers — you see the picture sharpen as you go.
03
Get your score
Your personal automation exposure profile. Which tasks are most exposed. Where your strongest human advantage is. How you compare to peers in your profession.
04
Share it
"Apparently only 23% of my job is actually automatable." A shareable card designed for the conversation everyone's already having.
Privacy
Anonymous by architecture,
not by policy.
We don't collect personal data because we don't need it. The system is designed so that identifying you is structurally impossible, not just prohibited.
✓
No login, no account, no cookies tracking identity
✓
No name, no email, no employer — your results are yours alone
✓
Your responses contribute to an anonymized public dataset — you can explore the results anytime
✓
One-click delete — everything is permanently removed on request
Full details: scoring model, synthetic data, and data use plans →
Why This Matters
The dataset that didn't
exist — until now.
Every automation forecast you've read is top-down — economists guessing from the outside. This is the first tool collecting bottom-up, task-level, worker-reported automation data.
As the dataset grows, it becomes a resource for AI companies understanding their real market, policymakers designing evidence-based workforce strategy, and universities building curricula for the skills that actually matter.
You get your personal profile. The world gets better data.
Explore
See what the data says.
The dataset is already large enough to surface patterns. What does genuine automation exposure look like across junior roles versus executives? Which professions are protected by physical presence alone? Where is the human advantage actually strongest?
This project runs on private savings. If you find it useful, a small contribution helps keep it going.