A readout, not a verdict
Is my job safe from AI?
Type a job title in plain language. We match it to the closest occupation category in Anthropic's own published AI-usage research and show you the real numbers — not a made-up safety score.
Enter a job title
What this actually shows
Anthropic — the company behind the Claude AI models — publishes the Anthropic Economic Index, a real research dataset on which O*NET/SOC occupation categories show up in actual Claude conversations, and how. It covers 756 occupations, scored on how much their tasks appear in measured AI usage, plus whether that usage leans toward the AI doing a task largely on its own (automation-style) or a person staying in the loop (augmentation-style).
This site does none of the scoring itself. It fuzzy-matches whatever job title you type to the closest occupation category, then shows you Anthropic's own published numbers for that category, plain and unedited. Everything runs in your browser — no account, no server, no tracking of what you search.
Frequently asked questions
Does this tell me if I'll lose my job to AI?
No. It shows how often tasks belonging to your closest-matching occupation category appear in real, measured Claude usage, and whether that usage leans toward automation-style or augmentation-style interactions. That's a snapshot of current usage patterns, not a forecast — Anthropic's own research explicitly frames it this way, and we do too.
Is this data specific to India?
No, and this is important. The underlying dataset is global, weighted toward US/English-language Claude usage, and organized around the US Department of Labor's O*NET occupation taxonomy — not Indian job titles or the Indian labour market. We map your input to the closest matching global category and say so clearly on every result.
Where does the data come from?
The Anthropic Economic Index (Hugging Face dataset repository), released under CC-BY-4.0. See the About & methodology page for exactly which files we use and how the automation/augmentation split is derived.
What's the difference between automation-leaning and augmentation-leaning?
Automation-leaning means the measured Claude conversations for that occupation's tasks tend to ask the AI to complete something largely on its own. Augmentation-leaning means a person stays more in the loop — iterating, validating, or learning together with the AI. Most occupations in the dataset show a mix of both.
Why didn't it find my exact job title?
The dataset covers 756 formal O*NET/SOC occupation categories, not every informal job title. We keep a large synonym list for common Indian phrasing (like "bank PO" or "SDE") and fall back to fuzzy text matching for anything else. If nothing matches confidently, you'll get the closest few categories to pick from instead of a dead end.
Do you store what I search?
No. Matching happens entirely in your browser using a data file shipped with the site — nothing you type is sent anywhere. See the privacy policy for the full picture, including hosting-level logs.