HomeResearch LibraryAccurate Failure Prediction in Agents Does Not Imply Ef…
Chapter 8 · 2026

Accurate Failure Prediction in Agents Does Not Imply Effective Failure Prevention

Rakshith Vasudev, Melisa Russak, Dan Bikel

Abstract

This paper investigates the effectiveness of LLM critic models in improving agent reliability. It demonstrates that while LLM critics can have high offline accuracy, their interventions at deployment time can lead to severe performance degradation. The study identifies a disruption-recovery tradeoff and proposes a pre-deployment test to estimate whether intervention is likely to help or harm, preventing regressions before deployment.

Eigenvector Insight — Zone III / PASF-PADE AnalysisNot part of the original paper
Eigenvector Research — Marco van Hurne
How this paper contributes to solving the Zone III problem (PASF-PADE)

This paper contributes useful building blocks for Zone III architecture through its work on LLM agents, failure prediction, failure prevention. While not exclusively focused on enterprise deployment, the insights translate directly to the challenges of long-horizon agentic workflows. The key lesson for Zone III practitioners: the problems identified here do not disappear at scale — they compound. Understanding them at the research level is prerequisite to solving them in production.

Why AI is not sufficient for Zone III without this

Zone III refers to high-complexity, high-risk, long-running agentic workflows — the class of enterprise AI deployments where a single failure can cascade across hundreds of steps. Standard AI models, trained to predict the next token, are not inherently designed for durable, governed, multi-step execution. This paper addresses one or more of the structural gaps that make Zone III deployments unsafe without explicit architectural intervention.

Topics

LLM agentsfailure predictionfailure preventionreliabilitycritic modelsdeploymentperformance degradation