India’s cities are growing faster than planning capacity, and as of 2024, 52% of statutory towns and 76% of census towns did not have a master plan to guide spatial growth and infrastructure investment. Early infrastructure ideas often begin as informal requests, but turning those ideas into useful planning questions can require specialized GIS tools, data, and expertise. My project: buildIN, is an open-source, India-first urban planning copilot for the first five minutes of a public-infrastructure proposal.
A user can ask: “Add a bus stop near Sector 17, Chandigarh.”

A local open-weight model, Phi-4-mini, interprets the proposal and identifies the intervention and location. buildIN then retrieves nearby evidence from OpenStreetMap, including existing bus stops, crossings, footpaths, major roads, schools, hospitals, shops, and marketplaces. Deterministic planning rules convert that evidence into access, safety, overlap, and demand flags. The model explains the result in plain language, while a grounding check rejects reports containing unsupported numbers. The project includes a local Ollama workflow, OpenStreetMap integration, and an MIT open-source license.
buildIN does not replace planners, statutory plans, or site surveys. However, it makes the first planning conversation faster, more transparent, and more accessible. Use cases include being used as an educational tool by architectural students, especially accompanied with on-ground work.