Laser: Governing Long-Horizon Agentic Search via Structured Protocol and Context Register
- Source: https://arxiv.org/abs/2512.20458
- Authors: Shuting Wang, Qiaolin Xia, Vich Wang, Herberttli, Bobsimons, Zhicheng Dou
- Published: 2025-12-23
- Category: Agentic Search
- Problem: Recent advances in Large Language Models (LLMs) and Large Reasoning Models (LRMs) have enabled agentic search systems that interleave multi-step reasoning with external tool use. However, existing frameworks largely rely on unstructured natural-language reasoning and accumulate raw intermediate traces in the context, which often leads to unstable reasoning trajectories, context overflow, and degraded performance on complex multi-hop queries.
- Approach: Laser defines a symbolic action protocol that organizes agent behaviors into three spaces: planning, task-solving, and retrospection. Each action is specified with explicit semantics and a deterministic execution format, enabling structured reasoning processes and reliable action parsing.
- Context Register: In coordination with parsable actions, Laser maintains a compact context register that stores only essential states of the reasoning process, allowing the agent to reason over long horizons without uncontrolled context expansion.
- Interpretability: The design makes intermediate decisions interpretable and traceable, enhancing explicit retrospection and fine-grained control over reasoning trajectories.
- Results: Experiments on Qwen2.5/3-series models across challenging multi-hop QA datasets show Laser consistently outperforms existing agentic search baselines under both prompting-only and fine-tuning settings.
- A Systematic Framework for Enterprise Knowledge Retrieval
- Retrieval Augmented Generation and Understanding in Vision: A Survey
- Synergizing RAG and Reasoning: A Systematic Review
- AceSearcher: Bootstrapping Reasoning and Search for LLMs via Reinforcement Learning
- Agentic Information Retrieval
Key Points
Original Abstract
> Recent advances in Large Language Models (LLMs) and Large Reasoning Models (LRMs) have enabled agentic search systems that interleave multi-step reasoning with external tool use. However, existing frameworks largely rely on unstructured natural-language reasoning and accumulate raw intermediate traces in the context, which often leads to unstable reasoning trajectories, context overflow, and degraded performance on complex multi-hop queries. In this study, we introduce Laser, a general framework for stabilizing and scaling agentic search. Laser defines a symbolic action protocol that organizes agent behaviors into three spaces: planning, task-solving, and retrospection. Each action is specified with explicit semantics and a deterministic execution format, enabling structured and logical reasoning processes and reliable action parsing. This design makes intermediate decisions interpretable and traceable, enhancing explicit retrospection and fine-grained control over reasoning trajectories. In coordination with parsable actions, Laser further maintains a compact context register that stores only essential states of the reasoning process, allowing the agent to reason over long horizons without uncontrolled context expansion. Experiments on Qwen2.5/3-series models across challenging multi-hop QA datasets show that Laser consistently outperforms existing agentic search baselines under both prompting-only and fine-tuning settings, demonstrating that Laser provides a principled and effective foundation for robust, scalable agentic search.
Context and Implications for Search / Rec / Personalization
1. Architecture: Cascaded retrieval + reranking + generation remains mainstream, but the agentic paradigm is making retrieval count and policy itself a learnable object. 2. Data: High-quality instruction data and click/session logs remain critical; synthetic data requires care against knowledge leakage and distribution shift. 3. Evaluation: The gap between offline metrics and online satisfaction is widening; LLM-as-judge should be cross-validated with human evaluation. 4. Deployment: Latency, cost, interpretability, and safety policies are hard constraints for industrial adoption—not just academic benchmark optimization.