GeoFlow: Agentic Workflow Automation for Geospatial Tasks
🤖 AI Summary
Existing intelligent agent workflows for geospatial tasks suffer from heavy reliance on manual design, low success rates, and excessive computational resource consumption. To address these challenges, this paper proposes a goal-driven automated workflow generation framework. Its core innovation lies in a fine-grained tool-call objective modeling mechanism: explicitly decomposing task-level semantic goals and jointly optimizing task planning with API invocation scheduling. The method leverages large language models (LLMs) to construct an agent architecture that integrates goal-guided task decomposition, tool selection, and runtime dynamic API scheduling. Evaluated on standard geospatial benchmarks, the approach achieves an average 6.8% improvement in task success rate across mainstream LLMs and reduces token consumption by up to 75%, significantly enhancing both the accuracy of complex geospatial reasoning and overall computational efficiency.
Application Category
📝 Abstract
We present GeoFlow, a method that automatically generates agentic workflows for geospatial tasks. Unlike prior work that focuses on reasoning decomposition and leaves API selection implicit, our method provides each agent with detailed tool-calling objectives to guide geospatial API invocation at runtime. GeoFlow increases agentic success by 6.8% and reduces token usage by up to fourfold across major LLM families compared to state-of-the-art approaches.
Problem
Research questions and friction points this paper is trying to address.
Automates geospatial task workflows with agentic design
Explicitly guides geospatial API selection via tool-calling objectives
Improves agent success rates while reducing computational costs
Innovation
Methods, ideas, or system contributions that make the work stand out.
Automatically generates agentic workflows
Provides detailed tool-calling objectives
Improves success and reduces token usage
🔎 Similar Papers
No similar papers found.
