Cotae

Existing plan-survey apps are capped where renovation needs them most — wall heights, free editing, and the insulation that decides a room's real surface.Existing plan-survey apps are capped where renovation needs them most — wall heights, free editing, and the insulation that decides a room's real surface.

Tools like MagicPlan are limited exactly where a renovation survey needs precision: a capped number of projects, imprecise wall heights, constrained manual editing, and — most of all — no notion of insulation, even though it's what determines a room's true surface after works. I needed a survey tool that produces exact, post-renovation surfaces I can trust, and that feeds straight into the Domus cost model rather than a separate silo.

A field tool that has to stay exact on site, talk to real hardware, and land its data in the same schema as the Domus calculator.A field tool that has to stay exact on site, talk to real hardware, and land its data in the same schema as the Domus calculator.

• Exactness on real sites. The surfaces drive real cost and yield estimates downstream, so a rough scan isn't enough — every wall height and area has to be right.
• Talk to the hardware I already use. A precision laser rangefinder is the ground truth for measurements, so the app has to read from it directly instead of me retyping numbers.
• One data model with the web app. Cotae isn't a standalone toy — its output has to land in the same Supabase / PostgreSQL schema the Domus calculator already reads.

Scan with LiDAR, refine at the laser, remodel freely — then offset insulation face by face for the real surface, all on one shared schema.Scan with LiDAR, refine at the laser, remodel freely — then offset insulation face by face for the real surface, all on one shared schema.

Cotae combines the iPad's LiDAR scan with a precision laser rangefinder and a bespoke plan editor. A RoomPlan / ARKit scan of the home as-is is converted into a clean, editable geometric model — free polygons, per-wall heights, sloped ceilings — rather than a locked mesh. Measurements are refined at the laser: I reverse-engineered the Bluetooth protocol of the Bosch UniversalDistance 40C (mapping the GATT service and decoding the BLE frames) so a reading drops straight into the app. From the scan you remodel freely — break walls, rearrange, adjust heights — and insulation is posed face by face, offset geometrically along polygon edges to give both the raw and the real post-works surface. Everything writes to a single Supabase / PostgreSQL schema shared with the Domus Next.js calculator, and exports structured data plus dimensioned PDF / JPG plans for the client.

An iPad app that scans in place, laser-refines, remodels to exact surfaces, and pushes structured rooms straight into Domus.An iPad app that scans in place, laser-refines, remodels to exact surfaces, and pushes structured rooms straight into Domus.

A working iPad surveying app: LiDAR capture of a home as-is, laser-accurate refinement over reverse-engineered Bluetooth, free remodelling with per-wall heights and sloped ceilings, and face-by-face insulation with raw and post-insulation surface areas. Its output feeds the Domus cost model over a shared database schema, and it exports dimensioned plans (PDF / JPG) for sharing with clients.

The editable geometry should have been the canonical model from the start, with the scan as just one importer feeding it.The editable geometry should have been the canonical model from the start, with the scan as just one importer feeding it.

The editable geometry grew out of the RoomPlan scan format, which quietly made the scan the source of truth. I'd invert that: define the canonical editable model first and treat RoomPlan as one importer among several (laser-only, fully manual), so the app doesn't lean on any single capture path. I'd also harden the Bluetooth layer — a reverse-engineered protocol is brittle to firmware changes, so a small capture-and-replay harness to re-derive frame formats quickly would save the next round of guessing.