Inspiration
After 22 years installing CCTV, access control, and structured cabling, I kept hitting the same wall: turning floor plans and site walkthroughs into reliable, itemized quotes still
took hours of manual counting and spreadsheet work. MCP VisionQuote was born from that friction — an AI estimator that combines Model Context Protocol (MCP/WebMCP) with domain
knowledge from real-world low-current projects. Every surveillance or cabling quote starts with the same questions: how many cameras, how much cable, which conduits, how many man-hours? Experienced integrators can estimate a site in
minutes, but clients still wait days for a polished proposal. I wanted to capture that expertise in a tool that reads project context through MCP tools, reasons about layouts and
requirements, and returns a quote an installer can trust.
What it does
How we built it
MCP VisionQuote exposes domain-specific tools through MCP/WebMCP: estimate_project, get_bom, generate_proposal, and suggest_camera_placement. The backend is Python 3.12 +
FastAPI, with a lightweight SQLite store and a seed price catalog. A FastAPI web UI lets users enter site parameters and instantly see a BOM, labor estimate, and commercial proposal. I
used YOLOv8-nano for optional photo-based placement suggestions, and Render + Cloudflare Tunnel for the public deployment.
Challenges we ran into
The hardest part was not the code — it was encoding field reality. Camera counts change when a hallway has no ceiling access. Cable lengths explode when plenum routes are blocked. I
had to balance AI speed with installer-grade accuracy, so every recommendation includes assumptions that a human can review and override. Trust, not full automation, became the real
goal.
Accomplishments that we're proud of
What we learned
Domain experience is the differentiator. Generic LLMs can write a quote; only a system grounded in low-current installation practice knows why a quote is wrong. MCP proved to be the right glue: it keeps project data separate from reasoning, making the estimator easier to audit, extend, and connect to other tools.
What's next for MCP VisionQuote
MCP VisionQuote is a practical step toward AI-assisted field estimating — built by someone who has pulled the cable, mounted the cameras, and argued about the labor hours.
Log in or sign up for Devpost to join the conversation.