AI & digital work
How much the job involves information, data, writing, analysis or other digital work that AI and software may help with.
How the tool works
How Toolsyte compares jobs
We look at information about what people actually do in hundreds of occupations. We consider how computer-based, repetitive and physically automatable the work is, along with how much it depends on human judgment, creativity and interaction. We then compare each occupation with other jobs.
This is a comparison tool, not a prediction that a job or person will be replaced.
What we compare
The tool brings four parts of work together to help explain the result.
Repeatable work
How much of the work follows regular, predictable or repeated steps.
Physical automation
How much of the physical work may be suitable for machines or robotics.
Human skills matter
How much the job depends on judgment, creativity, communication and adapting to real situations.
View full technical methodology
Approved Model B formula
Task/work-characteristic exposure
p = Physical Work Relevance / 100digital weight = 0.65 − 0.40probotics weight = 0.05 + 0.40ptechnical exposure = digital weight × Digital + 0.30 × Routine + robotics weight × Roboticsraw exposure = technical exposure × (1 − 0.25 × Human Reliance / 100)Physical Work Relevance is a methodology/routing dimension, not a public risk score. The formula is frozen as Model 2.1; occupation values are generated without manual overrides.
Relative Exposure Index
100 × (midrank − 1) / (N − 1)The public headline is a complete-universe percentile based on all 897 compatible occupations—not only the 89 launch pages. Exact raw-score ties use their midpoint rank. An index of 82 means higher modeled exposure than approximately 82% of occupations the current model can score. It is not an 82% probability of job loss, an automated-task percentage or a forecast of worker replacement.
Indices may change when O*NET data or the scorable occupation universe changes, even if one occupation’s raw value does not.
External Technology Signals
Some technologies replace or reorganize a workflow without appearing directly in the incumbent worker’s O*NET tasks, activities or tools. Where authoritative evidence exists, a separate External Technology Signal may appear. Signals are supplementary, non-numeric and never modify the Exposure Index. If no signal appears, that does not mean disruption is absent.
What the model does not forecast
Adoption and timing
The model does not forecast technology adoption, investment economics, employer decisions or timing.
Jobs and employment
It does not forecast job losses, employment decline, staffing levels or the share of tasks that will disappear.
Policy and constraints
It does not directly predict regulation, liability, infrastructure, public acceptance or deployment constraints.
Every external technology
The numerical model does not capture every customer-operated, autonomous or externally introduced substitution technology.
Coverage and exclusions
Generated metadata currently records 897 scorable occupations and 119 excluded occupations. Exclusions occur when required compatible O*NET ratings are unavailable. An excluded occupation remains searchable where published, but receives no proxy or invented score.
Model version 2.1. O*NET 31.0, released August 2026. Methodology updated 2026-09-20. Toolsyte created the model; the index is not an O*NET or U.S. government statistic.
This product includes information from the O*NET 31.0 Database by the U.S. Department of Labor, Employment and Training Administration (USDOL/ETA), used under the CC BY 4.0 license. Toolsyte has modified and scored some information; USDOL/ETA has not approved, endorsed, or tested these modifications. O*NET® is a trademark of USDOL/ETA.