I've created a tool based on the Anthropic Economic Index that lets you visualize AI capabilities across different occupations and tasks. Anthropic's data tracks actual usage by specific occupations on tasks in the O*NET database. Usage reflects actual adoption, but I wanted to understand current capability. To find this, I asked "where is AI being used for similar tasks in different occupations". I created vector embeddings for all task descriptions, found the nearest tasks semantically, and updated the capability score based on where AI is being used in comparable tasks across other occupations. I also did a sanity check with Claude Haiku on the scores to weed out cases where physical task rankings were inheriting scores from semantically similar but purely informational tasks. There are still places where this methodology doesn't capture everything (for instance, in the Web Developers occupation, many tasks like "Select programming languages..." are listed as 0% capability), but I'm trying to keep things grounded without too many judgment calls. Embedded widget should be visible in the post. If you'd like to see it on my website, here is the external link: How much of your job can AI handle? Thanks to Claude Opus 4.6 for helping with the programming here. Discuss