Semantic Context for Tool Orchestration
This work addresses the challenge of building sample-efficient, adaptive, and scalable orchestration agents for AI systems, representing a novel method for a known bottleneck rather than an incremental improvement.
The paper tackles the problem of tool orchestration by introducing Semantic Context (SC) as a foundational component, showing that SC-based methods like SC-LinUCB achieve lower regret in theoretical models and enable LLMs to effectively orchestrate over large action spaces, such as on a benchmark with over 10,000 tools.
This paper demonstrates that Semantic Context (SC), leveraging descriptive tool information, is a foundational component for robust tool orchestration. Our contributions are threefold. First, we provide a theoretical foundation using contextual bandits, introducing SC-LinUCB and proving it achieves lower regret and adapts favourably in dynamic action spaces. Second, we provide parallel empirical validation with Large Language Models, showing that SC is critical for successful in-context learning in both static (efficient learning) and non-stationary (robust adaptation) settings. Third, we propose the FiReAct pipeline, and demonstrate on a benchmark with over 10,000 tools that SC-based retrieval enables an LLM to effectively orchestrate over a large action space. These findings provide a comprehensive guide to building more sample-efficient, adaptive, and scalable orchestration agents.