how agent benchmarking efforts distribute across domains of human work
• Current benchmarks are concentrated in the Computer and Mathematical domain
• Leaving other highly-digitized domains (e.g., Legal, Management) underrepresented