Methodology

Most AI career tools ask what your job title is and spit out a number. That tells you almost nothing. Two marketing managers can have completely different risk profiles depending on which tasks fill their actual day.

We built this assessment around a simple idea: AI doesn't replace jobs. It replaces tasks. Your risk depends on which tasks you do, how much of your time goes to each one, and how far current AI capabilities have come for those specific activities.

The approach

We start with your role and pull the real tasks associated with it from federal occupational data covering over 900 job types. You tell us which of those tasks actually make up your work and roughly how much weight each one carries. That task selection is the foundation of everything that follows.

From there, we layer in three things:

Exposure. For each task, how capable are current AI systems at performing it? Not hypothetically, not in five years. Right now. We draw on published research measuring real-world AI performance across hundreds of task types, cross-referenced with economic data on where AI is actually being deployed.

Adoption. How much are you personally using AI in your work, and how does that compare to what we see from others in similar roles? Someone in a high-exposure role who's already working alongside AI tools faces a different situation than someone who hasn't started.

Resilience. Some parts of your work are harder to hand off than others. Interpersonal judgment, creative strategy, trust-based relationships, domain expertise built over years. We assess what makes your specific contribution difficult to replicate and factor that into your profile.

Two scores, one profile

Your results come down to two axes: how exposed your work is to AI automation, and how actively you're using AI in your role. Those two scores place you into one of eight career profiles, each representing a different combination of risk and readiness.

The profiles aren't labels meant to scare you. They're diagnostic. Someone with high exposure and high adoption is in a fundamentally different position than someone with high exposure and no adoption, even if their raw risk numbers look similar.

Why it works this way

Every design choice comes back to the same question: does this give someone a more accurate picture of where they stand?

We weight tasks by importance because not all tasks matter equally to your career. We factor in tenure and industry because the same role plays out differently in a 10-person startup versus a government agency. We include a self-assessment comparison because how people perceive their risk often diverges from what the data shows, and that gap is itself a risk signal.

The AI-generated section of your report is written by Claude (Anthropic's AI), but it doesn't calculate your scores. The scoring is deterministic. Claude's role is to translate your data into a narrative that connects your specific numbers to your specific situation.

What the data covers

Our task-level analysis draws on four published sources:

• O*NET (U.S. Department of Labor) for occupational task breakdowns and work activity classifications
• Anthropic Economic Index for measured AI capability across task categories
• Pew Research Center for workforce-level AI exposure frameworks
• "GPTs are GPTs" (Eloundou et al., Science 2024) for task-level automation exposure estimates

We don't use proprietary datasets. The research is published, peer-reviewed or government-maintained, and updated as new data becomes available. Our contribution is in how we combine these sources at the individual level, weighting and scoring them against your personal inputs.

What this is not

This is not a prediction. Nobody can tell you whether your specific job will exist in three years. Anyone claiming otherwise is selling something.

This is a snapshot. It measures where current AI capabilities intersect with your current work, based on the best available research. It identifies your blind spots, highlights where you're already strong, and gives you enough information to make better decisions about what to do next.

We update the underlying data and methodology as new research is published. The assessment you take today reflects what we know today.