Search papers, labs, and topics across Lattice.
The paper introduces INTENT, a novel inference-time planning framework for budget-constrained, tool-augmented LLMs that addresses the challenge of costly tool use in sequential decision-making tasks. INTENT employs an intention-aware hierarchical world model to predict future tool usage and risk-calibrated costs, enabling more effective online decision-making. Experiments on the cost-augmented StableToolBench demonstrate that INTENT significantly improves task success while strictly adhering to budget constraints and maintaining robustness under dynamic market conditions.
Forget brute-force search: this new framework lets LLMs plan tool use under strict budgets by reasoning about *intentions* and future costs, leading to much better task success.
We study budget-constrained tool-augmented agents, where a large language model must solve multi-step tasks by invoking external tools under a strict monetary budget. We formalize this setting as sequential decision making in context space with priced and stochastic tool executions, making direct planning intractable due to massive state-action spaces, high variance of outcomes and prohibitive exploration cost. To address these challenges, we propose INTENT, an inference-time planning framework that leverages an intention-aware hierarchical world model to anticipate future tool usage, risk-calibrated cost, and guide decisions online. Across cost-augmented StableToolBench, INTENT strictly enforces hard budget feasibility while substantially improving task success over baselines, and remains robust under dynamic market shifts such as tool price changes and varying budgets.
