Quoting for Geospatial AI

This experiment began as a Kaggle capstone, but it was built around a problem I knew from closer ground: how to turn complex technical requests into quotes that remain clear, structured, and defensible.

The workflow was designed for AI and geospatial projects, where the early stages are often shaped by ambiguity. Client needs had to be interpreted, internal services had to be matched with care, and the final proposal had to hold together as more than a rough estimate. To explore that, I built a multi-agent system that moved from conversation to requirements extraction, research, service recommendation, quote generation, review, and final Statement of Work.

What interested me most was not the novelty of using agents on their own, but the possibility of giving expert work a stronger frame. Not replacing judgment, but supporting it with structure. The result was a small but deliberate experiment in turning a messy professional workflow into something more consistent, auditable, and usable.