How Fast Do Agents Rot? An Empirical Study of Long-Horizon Degradation in LLM Agents for Production Decision-Making
The results argue for horizon-aware evaluation and reliability budgeting in place of aggregate pass-rate metrics, and for teams responsible for agent orchestration and reliability at scale to consider horizon-aware evaluation and reliability budgeting in place of aggregate pass-rate metrics.
Shubhra Mittal
· 1 citation