Search papers, labs, and topics across Lattice.
This paper introduces ToolPro, an innovative framework that transforms static web service interfaces into dynamic executable tool programs, enabling LLM-based agents to handle complex workflows with loops and conditionals. By employing constraint-guided program construction and effect-aware replay, ToolPro achieves significant reductions in end-to-end latency and client-side traffic, particularly in high-latency and complex scenarios. The evaluation demonstrates that ToolPro can reduce latency by up to 53.4% and traffic by up to 96.1%, marking a substantial improvement in the efficiency of agentic web service interactions.
ToolPro slashes web service latency by over 50% while drastically reducing client-side traffic, revolutionizing how LLMs interact with complex workflows.
In the agentic web era, LLM-based agents increasingly invoke web services as tools, yet most interfaces remain \emph{static endpoints} that poorly express long-horizon workflows with loops, conditionals, joins, and retries. We present ToolPro, which represents an agent's tool intent as an \emph{executable tool program} that compactly encodes multi-step service interactions with explicit effect types. ToolPro combines constraint-guided program construction, effect-aware replay for exactly-once state-modifying calls, and a profile-driven policy that decides when program execution outperforms stepwise calling. We instantiate ToolPro over MCP-style services with WebAssembly sandboxing and evaluate it on diverse workflows of real-world applications. ToolPro reduces end-to-end latency by up to 53.4\% and client-side traffic by up to 96.1\%, with larger gains under higher network latency and workflow complexity.
