GroundEval: A Deterministic Replacement for LLM-as-Judge in Stateful Agent Evaluation
2026-06-22 • Artificial Intelligence
Artificial IntelligenceComputation and LanguageSoftware Engineering
AI summaryⓘ
The authors present GroundEval, a method to check if an AI agent really used the right information to answer questions. Instead of just judging the final answer, GroundEval looks at the whole process the agent followed, including what it searched and cited. Their tests showed that sometimes agents give correct-sounding answers without actually retrieving the necessary evidence. GroundEval helps catch these hidden mistakes by examining if the agent truly based its answers on the right facts and reasoning steps.
GroundEvalAI agent evaluationevidence groundingLLM judgesaccess-controlled evidencecausal reasoningtrajectory analysisSilence trackPerspective trackCounterfactual track
Authors
Jeffrey Flynt
Abstract
Before letting an agent operate over real context, can you prove it used the right evidence? GroundEval turns that question into a deterministic test of what the agent searched, fetched, cited, and was permitted to access. In one case study, two frontier LLM judges scored a plausible agent response above 0.85. But the trace told a different story: the agent had never retrieved the artifact its answer depended on, yielding a GroundEval score of 0.000. We introduce GroundEval, a judge-free framework for evaluating agents against grounded, time-bounded, and access-controlled evidence. GroundEval uses a domain configuration to generate questions, lets the agent choose how to answer, and then scores both the final answer and the recorded trajectory that produced it. The benchmark targets three failures that LLM-as-judge evaluation struggles to detect: whether an agent checked before claiming absence, reasoned only from evidence available to the actor at the relevant time, and used the correct causal mechanism rather than a plausible one. These correspond to three tracks: Silence, Perspective, and Counterfactual. GroundEval exposes when plausible answers rest on invalid evidence paths, and produces structured per-question diagnostics that pair tool activity with the agent's turn-level narration, making each score inspectable rather than merely reported. What our case studies turned up is that this gap isn't some rare corner case. It's exactly the blind spot that final-answer and judge-based scoring were never built to catch.