40,000 jobs. Task by task. From the U.S. government.

The foundation of every tobywins.ai score is the Occupational Information Network (O*NET), maintained by the U.S. Department of Labor and updated continuously. O*NET breaks down 923 occupations into their specific tasks, knowledge requirements, and skill demands — covering over 40,000 distinct job variations.

When you type a job title, Toby matches it to the correct O*NET occupational category and scores each task individually based on how automatable that specific task is. The result is a score that reflects what the work actually requires — not just what the job title sounds like.

O*NET
O*NET OnLine — U.S. Department of Labor
Primary source for all occupational task data. Updated quarterly. Covers 923 occupational categories and more than 40,000 job variations with task-level descriptions, skill requirements, and employment data. Public domain — no copyright restrictions.
onetonline.org
BLS
U.S. Bureau of Labor Statistics — Occupational Outlook Handbook
U.S. government employment projections by occupation through 2033. Used to validate directional risk signals — roles with declining employment projections score higher risk, roles with growing projections score lower risk. Public domain.
bls.gov/ooh

Task-level scoring, not title-level guessing

Most job risk tools give you a single number based on your job title alone. Toby scores each task inside your role separately, then combines them into an overall protection score from 0 to 100. Higher means more protected. Lower means more of the role can be automated today.

This matters because two people with the same job title can have very different risk profiles. A lawyer who primarily writes routine contracts is more exposed than one who handles complex negotiations. A nurse who primarily documents and follows checklists is more exposed than one who builds patient relationships and makes clinical judgment calls in complex situations.

Factor Direction Research basis
Task automation probability Primary factor O*NET + Frey & Osborne methodology
Physical presence required Lowers risk MIT Iceberg Index, 2025
Social and emotional intelligence Lowers risk OpenAI/UPenn, 2023
Judgment under uncertainty Lowers risk McKinsey Global Institute, 2025
Routine pattern-following Raises risk OpenAI/UPenn, 2023
BLS employment trend Directional signal U.S. Bureau of Labor Statistics, 2025
What the score does not tell you

Your score reflects the automation risk of your job category based on its typical task mix. Two people with the same title at different companies can have meaningfully different real risk levels depending on how they actually spend their time. The score is a starting point for understanding your exposure — not a prediction of whether you personally will lose your job.

The studies behind every score

Toby's scoring model is validated against six peer-reviewed and government research sources. All are open access or freely citable. Sources listed most recent first.

Stanford Digital Economy Lab · August 2025
Canaries in the Coal Mine? Entry-Level Employment and AI
The first large-scale empirical study of actual AI-driven job displacement. Using ADP payroll data from millions of workers, found entry-level employment for workers aged 22 to 25 in software development is down 20% from its 2022 peak. Customer service down 11%.
Stanford Digital Economy Lab
MIT & Oak Ridge National Laboratory · November 2025
The Iceberg Index: Skills-centered AI Exposure in the U.S. Economy
Simulates 151 million workers across all 50 states. Finds AI can technically perform tasks equivalent to 11.7% of the U.S. workforce — with hidden exposure in cognitive and administrative roles 5 times larger than visible tech-sector disruption.
arxiv.org/html/2510.25137v2
World Economic Forum · January 2025
Future of Jobs Report 2025
Projects 170 million new roles created and 92 million displaced globally by 2030. Based on survey of 1,000+ employers representing 14 million workers across 22 industries and 55 countries. Healthcare, education, and advisory roles projected to grow 15 to 25%.
weforum.org
McKinsey Global Institute · 2025
AI Adoption and Workforce Transition
Less than 5% of occupations can be fully automated with current technology, but 60% have significant partial automation exposure in specific tasks. Entry-level workers face transition risk 14 times higher than experienced workers in the same role.
McKinsey Global Institute
OpenAI & University of Pennsylvania · March 2023
GPTs are GPTs: Labor Market Impact of Large Language Models
80% of U.S. workers have at least 10% of their tasks affected by large language models. 19% have more than 50% of tasks affected. Uses the same O*NET task database as tobywins.ai. Published on arXiv, open access.
arxiv.org/abs/2303.10130
Oxford University · 2013 · Foundational methodology
The Future of Employment: How Susceptible Are Jobs to Computerisation?
The paper that established the task-level scoring methodology now used across this entire field — including by MIT, OpenAI, and tobywins.ai. Published as a working paper in September 2013. Specific risk percentages are superseded by more recent research, but the analytical framework it introduced remains the foundation of every serious job risk model.
Oxford Martin School

See how your job scores

Free, task by task, based on government data and peer-reviewed research. Takes 60 seconds.

Check your job free at tobywins.ai