Legal AI agents improve decisions by evolving during use
Test-Time Agent Evolution for Long-Horizon Legal Reasoning
Artificial Intelligence
Summary
Legal decisions often involve many steps and different roles, which can change as new facts and evidence come in. The authors studied how AI agents can get better at helping with these steps by learning from past cases during use without changing their core programming. Their approach lets agents remember past experiences and check each other’s actions to work together better. Tests showed this method helped AI agents make more reliable legal decisions over long processes.
What this means in practice
- •For legal technology developers: Build AI assistants that adapt during deployment to improve legal process decision-making across multiple roles and stages.
- •For business process automation teams: Create workflows that use evolving AI agents to handle complex, multi-step decision tasks in dynamic environments.
Authors
Haotian Chen, Shuaicheng Niu, Haocong Rao, Kaisong Song, Jun Lin, Lizhen Cui, Zhiqi Shen, Yonghui Xu
Abstract
Legal intelligence aims to support reliable decision-making across long-horizon legal processes involving evolving case states and multiple roles. However, real-world legal deployment exhibits substantial case heterogeneity in facts, evidence, and procedural contexts, exposing the limitations of static agent strategies. Moreover, legal reasoning is inherently interdependent across roles and procedural stages, making global reliability fundamentally different from isolated role competence. To address these challenges, we study training-free test-time agent adaptation, where agents continuously exploit deployment-time signals from preceding cases and ongoing interactions without updating model parameters. We propose \method, which introduces \emph{Test-Time Memory Evolution} to retrieve reusable experience from previous cases, adapt it to the current factual and procedural context, and consolidate accumulated experience for subsequent decision-making. Further, \emph{Rubric-Aligned Collaboration} verifies and revises role-specific actions according to behavioral and procedural requirements, enabling coordinated decision-making across roles and stages. Extensive experiments on J1-EVAL and LegalWorld across five backbone models demonstrate consistent improvements over representative reasoning and agent baselines with reasonable interaction and computational costs. Ablation and case studies further show that the two components provide complementary benefits in experience adaptation and cross-role coordination, improving the reliability and efficiency of long-horizon legal reasoning.