Small language models improve checking AI tool choices for task relevance
Toward SLM-based agentic task-tool intent matching
Artificial Intelligence
Summary
AI agents often use different tools to complete tasks, but sometimes they pick tools that don't really help with the main goal. The authors studied if smaller AI language models can check each tool choice quickly to see if it's actually helpful for the task. They created a new dataset where tasks need multiple tools and trained these small models to judge tool relevance better. This can help catch mistakes or bad choices while the AI is working.
What this means in practice
- •For ai system developers: Add efficient real-time checks for AI tool calls to ensure they logically align with task goals, improving reliability on local or private setups.
- •For automation platform engineers: Integrate small language model based tool relevance verification to detect and prevent unintended or harmful tool usage in multi-agent workflows.
Authors
Chiara Troiani, Arash Salarian, Majed El Helou, Benjamin Ryder, Jean Diaconu, Hervé Muyal, Marcelo Yannuzzi
Abstract
Tool-equipped AI agents use tool calls to access data and act on external systems. Horizontal growth of agentic systems increases the number of these interactions, and further motivates the need for automated, per-call oversight that can operate at low latency and/or on-prem. Conventional authorization schemes can determine whether an agent is allowed to invoke a tool, but cannot assess the agent's underlying cognition, specifically, whether the tool selection represents a logical, relevant step toward satisfying the intent of the task or not. Consequently, an allowed call may still deviate from the task's intent: a rogue agent might deviate the calls or nudge other agents to make a combination of calls that would not align with the intent of the task. Therefore, every call needs to be verified. In this study we investigate the applicability of Small Language Models (SLMs) to this purpose: an SLM functions as a task-tool relevance classifier that evaluates every selected tool independently against the assigned task and returns a relevance signal for downstream enforcement. Equipped with a novel dataset with multi-tool tasks whose required tools span distinct Model Context Protocol (MCP) servers, we used prompt-optimization, supervised fine-tuning, and reinforcement learning through GRPO to optimize and specialize SLMs.