InsightEmb: Learning Action-Intent Embeddings for Agentic Insight Retrieval
This work addresses the problem of retrieving actionable insights for self-improving agents, offering a method that transfers across domains and reduces the need for expensive environment-specific supervision.
The paper introduces InsightEmb, a contrastive embedding framework for agentic insight retrieval that learns progress-oriented retrieval geometry from mathematical reasoning data. It outperforms existing reasoning embedding models on dynamic agent tasks and a static skill-retrieval benchmark without environment-specific training.
Self-improving agents accumulate reusable insights from prior trajectories, making retrieval increasingly important for turning accumulated experience into actionable guidance. At each decision step, retrieving the right insight can help the agent progress toward its goal, a setting we refer to as agentic insight retrieval. However, existing retrieval methods primarily model semantic similarity, while overlooking whether a retrieved insight resolves the agent's current decision bottleneck. We propose InsightEmb, a contrastive embedding framework that learns transferable progress-oriented retrieval geometry using only mathematical reasoning data. InsightEmb jointly learns to align concrete situations with abstract heuristic rules and to cluster reasoning trajectories with similar progress structures. We evaluate InsightEmb on dynamic agent tasks and a static skill-retrieval benchmark. Without any environment-specific training, InsightEmb improves over all these evaluations, surpassing the performance of existing reasoning embedding models. These results suggest that the geometry of state-insight matching can transfer across domains, enabling effective training from publicly available reasoning data without expensive environment-specific supervision.