RL-ADA: A World-Feedback Framework for Adversarially Robust Enterprise Dialogue Agents

Researchers introduced RL-ADA, a co-evolutionary framework using world feedback instead of human labels to train adversarial robust enterprise dialogue agents, eliminating the annotation bottleneck. In a banking proof of concept, the framework doubled the strict end-to-end PASS rate over five cycles and eliminated tool-routing errors, while also revealing an emergent adversarial strategy called Contextual Camouflage.

arXiv:2609.02902v1 Announce Type: new Abstract: Deploying task-oriented dialogue agents in enterprise customer support faces a persistent annotation bottleneck: robust training requires labelled interaction data at scale, yet enterprise conversational logs are privacy-sensitive and expensive to annotate, while user behaviour evolves faster than labelling pipelines can keep pace. We present RL-ADA Reinforcement Learning with Adversarial Dialogue Agents , a co-evolutionary training framework that eliminates this bottleneck by replacing human labels with \emph{world feedback}: consequence-based reward signals derived directly from measurable interaction outcomes. A Customer Support Agent DA, 3B parameters and an Adversarial Customer Agent CA, 7B parameters co-evolve in an adversarial arena guided by a fixed automated judge: the DA is rewarded for correctly handling multi-turn customer conversations to successful resolution, while the CA is rewarded for producing realistic, intent-concealing utterances that cause misroutes, creating asymmetric adversarial pressure through opposing but independently structured rewards. An isolation gym iteratively retrains the weaker agent on prior-failure transcripts, requiring no human annotation at any stage. In a banking customer support proof of concept, tool-routing errors are eliminated and the strict end-to-end PASS rate doubles over five co-evolutionary cycles, driven solely by automated arena reward with no labelled data. We additionally observe the emergence of \textbf{Contextual Camouflage}, an adversarial strategy in which the CA learns to embed intent within dense realistic customer detail purely from reward pressure, with direct implications for enterprise red-teaming and robustness evaluation.