Current AI models achieve only 0.
FDE-Gym is a benchmark of 95 real-world delivery tasks where AI agents must clarify ambiguous customer requirements, work in live systems, and discover unstated constraints—mirroring the work of forward-deployed engineers.