Laser: Governing Long-Horizon Agentic Search via Structured Protocol and Context Register
Paper: https://arxiv.org/abs/2512.20458 Authors: Shuting Wang, Qiaolin Xia, Vich Wang, Herberttli, Bobsimons, Zhicheng Dou Published: 2025-12-23 Category: Agentic Search
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.
Key Points
- Problem: Agentic search systems that interleave multi-step reasoning with tool use suffer from unstructured natural-language reasoning, raw trace accumulation in context, unstable trajectories, context overflow, and degraded multi-hop QA performance.
- Solution — Symbolic Action Protocol: Laser organizes agent behaviors into three spaces (planning, task-solving, retrospection), with each action defined by explicit semantics and a deterministic execution format for reliable parsing and traceable intermediate decisions.
- Solution — Context Register: A compact register stores only essential reasoning states, preventing uncontrolled context expansion during long-horizon reasoning.
- Results: Evaluated on Qwen2.5/3-series models across challenging multi-hop QA datasets, Laser consistently beats existing agentic search baselines in 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
Background
Agentic search in large-scale search, recommendation, and personalization systems faces persistent challenges in efficiency, scalability, and intent understanding. Traditional pipelines split retrieval, ranking, and generation, which struggles to meet LLM-era demands for natural-language interaction, multi-hop reasoning, and real-time knowledge. Laser targets this intersection, rethinking how retrieval frequency and strategy become learnable decision objects rather than fixed pipeline stages.
Significance for Search / Rec / Personalization
1. Architecture: Cascaded retrieval + rerank + generation remains mainstream, but the agentic paradigm makes retrieval count and policy themselves learnable. 2. Data: High-quality instruction data and click/session logs remain critical; synthetic data must guard against knowledge leakage and distribution shift. 3. Evaluation: The gap between offline metrics and online satisfaction is widening; LLM-as-judge needs cross-validation with human assessment. 4. Deployment: Latency, cost, interpretability, and safety are hard constraints for industrial adoption — academic benchmarks alone are insufficient.
Limitations and Future Directions
Potential limitations include experiment scale bounded by GPU budgets, benchmark-vs-real-user distribution mismatch, unknown cross-lingual generalization from English-centric data, and safety risks of open-web agent operation. Future work may explore more efficient test-time compute allocation, deeper integration with knowledge graphs and structured databases, and causal/fairness constraints for recommendation systems.