Study Tests AI Agents on 50 Professional Tax Assignments

A new evaluation published on Surge HQ's blog assesses how AI agents perform on professional tax work. The benchmark, called Dayjob: Tax, consists of 50 assignments modeled on real tax tasks. The article is shared on Hacker News, where it drew a modest level of attention with 12 points and 6 comments. The source text provided is limited to the headline and metadata, so the specific results, methodology details, and agent performance figures are not described here.