Dayjob: Tax – evaluating AI agents on 50 professional tax assignments
Researchers at Dayjob: Tax tested AI agents by assigning them 50 professional tax tasks, such as filing returns and interpreting regulations. The study measured accuracy, speed, and compliance with legal standards. Results showed varying performance, with some agents completing tasks quickly but missing nuances, while others were slower but more precise. The work highlights challenges in automating complex tax workflows and suggests areas for improvement.