From the course: Using Spatial Data in AI Workflows
Join today to access over 25,600 courses taught by industry experts.
Design location‑to‑language agentic workflows
From the course: Using Spatial Data in AI Workflows
Design location‑to‑language agentic workflows
When we built our MCP server, we built a passive tool. It waited for us to ask a question, did one specific task, and returned an answer. But what if we have a complex problem? What if we want to say, find the center of Portland, then look for retail locations within 500 meters, count them, and tell me if that density is high or low. A data retrieval tool like an MCP cannot do that. For that, we need an agent. In this section, we are going to build a geospatial AI agent. This is a system that can reason, plan, and execute multiple steps to solve a spatial problem. Think about it like a workflow rather than a question and answer session. Let's look at the architecture we're about to implement. At the core, we have the brain. This is our large language model. In this project, we are using the open-source Llama model from Meta via LangChain. The problem is that the brain doesn't know about geometry, but it knows how to plan. Surrounding the brain, we have tools. In our tools.py file, we…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.