# Victorian planning answers, with the rule behind every one.

Source: https://cruxlabs.com.au/case-studies/propychat

Client: Our in-house startup · Sector: Proptech · Services: AI agents, MVP, Custom software, UI/UX design · Status: Early access

1. Every site opens with its zone, schedule and overlays.
2. Ask in plain words. Each answer shows the planning rule it came from.
3. It measures the lot on the map and tests what fits against the zone's own rules.

## The challenge

Planning answers take hours to days, and the data behind them is public but scattered.

## What we built

A public AI chat with the planning rule cited on every answer, property cards, site analysis, a property finder, suburb intelligence, a developer API, and the state's planning data held in our own database.

## How the automation works

Ask in plain words. The system finds the property, reads the zone, schedule and overlays, measures the lot and tests what fits against the zone's own rules. When the rules are silent, it says so.

## What changed

- **4.1M+**: Victorian properties, with their planning, that PropyChat answers about. Measured: count of properties in the PropyChat database, 14 September 2026 (4,187,991).
- **9 months**: first line of code to early access.

## Stack

Real code, Postgres with PostGIS and 33 GB of planning data, hosted in Sydney, the Claude AI model.

## Related work

- [The chauffeur business we scaled](https://cruxlabs.com.au/case-studies/chauffeur-business-system)
- [Crux Practice: practice management for accounting firms](https://cruxlabs.com.au/case-studies/crux-practice)
- [AI agents](https://cruxlabs.com.au/services/ai-automation)
- [Property](https://cruxlabs.com.au/industries/property-real-estate)
- [Visit PropyChat](https://propychat.com.au)

[Book a call](https://cruxlabs.com.au/contact) · [All our work](https://cruxlabs.com.au/case-studies)