· 1 min read
How to Use ChatGPT for Route Planning
What ChatGPT can and cannot do for route planning, and how connecting it to an optimization engine through MCP gives reliable results.
People often ask ChatGPT to "plan my route" and get a confident answer. For a few stops that may be fine. For a real delivery day it is risky.
What ChatGPT is good for
- Brainstorming how to group stops
- Explaining what a time window or capacity constraint does
- Writing driver notes or customer ETAs from a finished plan
Why it struggles with real routing
- No guaranteed optimization. A model predicts text; it does not search the solution space.
- Constraint blindness. Capacity and time windows are easy to violate unnoticed.
- No reproducibility. Ask twice, get two different plans.
- Data exposure. Pasting a customer list into a chat shares more than necessary.
Connecting ChatGPT to an optimizer
The Model Context Protocol (MCP) lets an assistant call tools on a server. The assistant asks the server to validate an import or run an optimization, and receives structured results. With AIRO, the engine does the maths and ChatGPT presents the outcome.
Whether MCP connectors are available in your ChatGPT plan depends on OpenAI's rollout, so verify that before planning around it. The web portal provides the same workflow without any assistant.
A reliable workflow
- Upload the spreadsheet through a secure link to the platform
- Ask the assistant to validate it and list problems
- Fix issues, then request an optimization for your vehicles
- Review KPIs and unassigned stops
- Export per-driver files
Join early access
Join the AIRO waitlist and tell us you use ChatGPT.