The Art of Tool Interface Design
This addresses the challenge of complex reasoning in customer service for businesses, but it is incremental as it builds on a standard agentic architecture with innovations in tool interface design.
The paper tackles the problem of improving reasoning capabilities in customer service scenarios by introducing the Thinker framework, which achieves state-of-the-art performance with success rates of 82.6% on GPT-4o and 81.9% on Llama-3.1 405B on the τ-bench retail dataset.
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.