Work Catalog & Estimate Items
Five work groups, 27 subtypes — each with a price range per unit, difficulty, time estimate, and visit flags. One estimate holds several items: tiling, painting, and baseboards in a single quote.
CASE STUDY — CONSTRUCTION
A small renovation crew wanted to cut the time a client waits while a master figures out the price. The tool lets the operator quote a realistic range in one minute — with AI filling the form straight from client photos.
HOW IT WAS BUILT
ClientRenovation crew of 4 masters
DomainConstruction & Renovation
ServicesRequirements · Logic · Web app · AI integrations
StackPython · Vue 3 · Claude API · pytest
Five work groups, 27 subtypes — each with a price range per unit, difficulty, time estimate, and visit flags. One estimate holds several items: tiling, painting, and baseboards in a single quote.
The operator drops the photos a client sent in the messenger — Claude vision detects work types, estimates quantities, and spots complications like mold or crumbling plaster. Every AI-filled field carries a confidence badge, and the form is never locked.
Ten complication codes add percentages to the whole estimate and raise difficulty; a floor price protects against underquoting. Visit-flagged work, critical codes, or totals over $2,500 trigger a visible escalation banner.
The tool writes the answer for the messenger: price range, recommended figure, duration for the chosen crew size, a labor-only note, and the free on-site visit disclaimer. From call to reply — one minute.
Under the hood: a pure, fully unit-tested calculation engine — wrapped in a Python app, with a Claude-powered photo-analysis endpoint that degrades gracefully to manual input. The same engine can be exposed as MCP tools, and the whole catalog is editable per client — the first customization point when selling the tool.
Python+
Vue 3+
Claude API+
Vision+
xUnit+
Estimates+
MCP+
Construction+