Recently, researchers have explored leveraging the commonsense knowledge and reasoning capabilities of large language models to accelerate human mobility simulation. However, these methods suffer from several critical shortcomings, including inadequate modeling of urban spaces, poor integration with individual mobility patterns, and weak alignment with collective mobility distributions. To address these limitations, we propose Geospatial Knowledge Enhanced Agentic framework for Mobility Simulation (GAMS), an agentic framework that leverages the language based geospatial foundation model to simulate human mobility in urban space. GAMS comprises three core modules: MobExtractor, which extracts template-based mobility profiles and synthesizes new ones during simulation; GeoGenerator, which generates trajectory points considering collective mobility knowledge and urban geospatial knowledge; and TrajEnhancer, which captures individual mobility regularities from real-world trajectories and refines to generate style-aligned synthetic trajectories. Experiments on two real-world datasets show that GAMS achieves superior performance over various state-of-the-art methods, with more than a 17% improvement by capturing diverse mobility regularities. Our codes and datasets are open-sourced via https://anonymous.4open.science/r/CAMS-1992/.

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 simple and 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.