JW Tech
All ventures
L3 · AI-Native Operations·In development

TerraLogic

Property feasibility analysis on Australian open data

The idea

Working out whether a development site is worth buying is slow, expert work: pull the zoning, check the overlays and hazards, estimate the yield, stack up the costs, and land on what the site is actually worth. TerraLogic does that automatically. You give it an address; it gives you a feasibility report — built end to end on Australian government open data, so the inputs are authoritative and free.

How it works

An address is geocoded and spatially joined against a stack of NSW government datasets — valuation, cadastre, the planning API, transit, and census / socioeconomic layers. A deterministic model then runs the numbers: yield, a cost stack (construction, contributions, stamp duty, acquisition), and the residual land value. Each site gets a Development Potential Score across planning capacity, risk, transit access, and demand, and the whole thing is written up as an AI-narrated report in Markdown, HTML, or PDF.

Architecture

The engine is Python and FastAPI over PostGIS, with GeoPandas and DuckDB for the spatial and analytical work and an XGBoost model for price prediction. It exposes two surfaces — a Next.js product UI and a versioned public API — over a medallion-style ETL pipeline designed to scale onto AWS for production.

Status

In active development. The data pipeline, feasibility models, and report generation are working; the focus now is hardening the product surfaces and the production data infrastructure.

Ready to Start

Let's Build Something
That Works at Scale

Wherever you are on the path — building a digital foundation, automating operational work, or putting AI into production — we'd like to understand the problem first.

No commitment required. We start with a discovery conversation to understand if there's a fit.