GroundCut + Mapbox

No. GroundCut maintains the Mapbox account at the platform level. All Mapbox API calls — geocoding, satellite tiles, Directions — run on our infrastructure. You don’t pay Mapbox directly; that cost is included in your $99/mo subscription. If you’re a heavy user approaching platform-level rate limits we’ll work with you on it.

Parcel data tells you the total lot size, including the house, driveway, garden beds, fenced patio, and shed — none of which you mow. The mowable area is what you need to price. Satellite imagery makes the customer pause, look at their actual yard, and outline what you’d cut. The drawn area runs through @turf/turf to compute square footage, which feeds the price-per-quarter-acre rule. Empirically, customer-traced quotes are far closer to walkthrough quotes than parcel-derived ones.

Within a few percent for residential lots up to about an acre. The customer’s tap precision is the dominant error term — mapbox-gl-draw lets them undo, redo, and redraw, which solves most of it. The pricing rules cap downside via min_price (you never quote below your floor) and max_price (you never quote above a sanity ceiling no matter how big the drawing is).

Mapbox refreshes its imagery on a regional schedule. Recently developed neighborhoods sometimes show a vacant lot when a house exists. The drawing tool works regardless — the customer outlines what they want mowed today, even if the imagery shows yesterday’s ground cover. New construction is the only place this is a regular issue.

No, deliberately. We use nearest-neighbor over a Mapbox Directions distance matrix starting from your home base. For a residential mowing day with 6–12 stops, nearest-neighbor produces routes within a few percent of the optimum and runs in under a second. Genuine TSP solvers would add latency without meaningful daily-time savings at the route sizes lawn crews actually run.