A feasibility workspace where every zoning rule traces back to its document

It turns site documents into a costed development option. In TerraScope AI, zoning rules are extracted from uploaded PDFs, checked by a person, and then used to shape a design brief, several scored concepts and a cost estimate. The final feasibility report keeps a link from each constraint back to the document that set it.

It can extract them, but a person should confirm them before they are used. Here each document is processed chunk by chunk, and every chunk gets a verify or reject decision. Height limits, coverage, setbacks and floor area ratio are shown with the source snippet and a confidence score, so the reviewer can check them quickly.

Yes. Stand schedules are extracted into structured rows, with the original snippet attached to each one. That turns a scanned table into data the brief, concepts and costing steps can calculate against, while still letting a reviewer see exactly which part of the document each stand area came from.

Because a guessed construction cost is worse than no cost at all. The estimate is calculated from base area, a rate per square metre and risk multipliers, then shown as a range in rand with a confidence percentage. AI can explain the result, but the numbers come from a fixed formula that gives the same answer every time.

The estimate says so directly. If the brief has no budget cap or there is no local contractor rate source, those gaps are listed as missing inputs and assumptions on the cost card. The platform does not fill them in quietly, so the reader knows how much weight the estimate can carry.

Each concept is scored on the same five criteria: privacy, view quality, buildability, cost efficiency and compliance confidence. Generating several scored options, instead of one answer, makes the trade-offs visible. A concept that responds well to a slope or keeps the best views can be weighed against one that is cheaper to build.