ReAct

“ReAct~Yao2023ReAct, Auto-GPT~Richards2023AutoGPT, and GAIA~mialon2023gaia explored the interaction between reasoning and acting, though often in synthetic or text-only environments”

“ReAct combines reasoning with API calls for multi-step tasks, but its performance is constrained by pretraining and degrades with increased tool complexity.”

“These results imply that curated data and retrieval augmentation, not sheer parameter count, are the present keys to dependable LLM tool use.”

“they operate at the agent level rather than providing fine-grained reasoning for individual function parameters”

“The dominant approach for orchestrating tool usage relies on reactive, step-by-step reasoning frameworks like ReAct, often augmented by self-reflection techniques. However, this paradigm suffers from inherent local optimization traps due to its incremental decision-making process. While potentially effective for simple queries, its reactive nature often falters on complex tasks.”

“These studies substantially advance reasoning control and tool-use alignment, but they generally treat reasoning depth and execution structure as separate concerns rather than as jointly case-conditioned aspects of the same problem.”