From the course: Using Spatial Data in AI Workflows
Join today to access over 25,600 courses taught by industry experts.
Decision framework for selecting spatial AI architectures in projects
From the course: Using Spatial Data in AI Workflows
Decision framework for selecting spatial AI architectures in projects
We have defined our three patterns, MCP, Agents, and Apps. But if you choose the wrong one, you'll waste time and tokens. You don't need a complex autonomous agent just to look up a zip code. To choose the right tool, we're going to use a simple decision tree based on two questions. Question one, what is the primary interface? If your users need to explore data, toggle layers, or understand spatial relationships visually, stop there. You need a map application. Chat windows are not able to convey spatial context and arrangement. If the output is, where are all the flood zones, show them a map. If the interface is text, like a chatbot or an API, then ask question two. Is the task retrieval or reasoning? If the user is asking for existing facts, like how many stores are within two kilometers of Shibuya Crossing, or how many parks are in Buenos Aires, then use MCP. MCP is fast, cheap, and deterministic. It connects the LLM to the database to fetch the truth. However, if the task requires…
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.