mlsanityData to 7 Oct 2026

The Art of Tool Interface Design

arXiv 2503.21036 ↗First posted 26 Mar 2025cs.AIcited by 0

On the map: LLM agents and tool use → Tool learning and function calling

Abstract (arXiv)

We present an agentic framework, Thinker, which achieves state of art performance in challenging reasoning tasks for realistic customer service scenarios that involve complex business logic and human interactions via long horizons. On the τ-bench retail dataset, Thinker achieves 82.6% success rate with GPT-4o (version 2024-06-01) (baseline: 68.3%), and 81.9% success rate with Llama-3.1 405B (baseline: 49.6%), without any fine-tuning. Thinker effectively closes the gap in reasoning capabilities between the base models by introducing proper structure. The key features of the Thinker framework are: (1) State-Machine Augmented Generation (SMAG), which represents business logic as state machines and the LLM uses state machines as tools. (2) Delegation of tasks from the main reasoning loop to LLM-powered tools. (3) Adaptive context management. Our prompting-only solution achieves signficant gains, while still maintaining a standard agentic architecture with a ReAct style reasoning loop. The key is to innovate on the tool interface design, as exemplified by SMAG and the LLM-powered tools.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Tool learning and function calling, one of 16 directions in LLM agents and tool use. Builds on 6 of its 14 references that are in the map. First posted on arXiv 26 Mar 2025.

Builds on

6 of 14 references.

References · 6

  1. 2023MemGPT: Towards LLMs as Operating Systemscited by 1,295
  2. 2024τ-bench: A Benchmark for Tool-Agent-User Interaction in Real-World Domainscited by 1,280
  3. 2024Long-context LLMs Struggle with Long In-context Learningcited by 285
  4. 2023ADaPT: As-Needed Decomposition and Planning with Language Modelscited by 172
  5. 2024StateFlow: Enhancing LLM Task-Solving through State-Driven Workflowscited by 65
  6. 2024Sparse Rewards Can Self-Train Dialogue Agentscited by 6

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026One Success Isn't Reliability: Thinkingbox, a Sandbox and Benchmark for Agents in Stateful Business Workflows
  2. 2025MindFlow: Revolutionizing E-commerce Customer Support with Multimodal LLM Agents
  3. 2026FAMA: Failure-Aware Meta-Agentic Framework for Open-Source LLMs in Interactive Tool Use Environments
  4. 2026Salesforce Koa: An Enterprise Language Model for Agentic Tool Use
  5. 2026RetailBench: Benchmarking long horizon reasoning and coherent decision making of LLM agents in realistic retail environments
  6. 2026AgentEscapeBench: Evaluating Out-of-Domain Tool-Grounded Reasoning in LLM Agents
  7. 2026EComStage: Stage-wise and Orientation-specific Benchmarking for Large Language Models in E-commerce
  8. 2026The Evolution of Tool Use in LLM Agents: From Single-Tool Call to Multi-Tool Orchestration