State of AI in Australian Property Technology (2026)

The Agentic Shift in Property

In January 2026, ATTOM—a major US property data provider—launched a Model Context Protocol (MCP) server with 55 tools for AI agents to access property valuations, comparable sales, and transaction history. It was the first major property data company to make its data directly accessible to AI agents through a standardised protocol.

MCP is a protocol developed by Anthropic that standardises how AI agents connect to external data sources. Think of it as a USB port for AI: instead of building custom integrations for every AI platform, a data provider builds one MCP server and every MCP-compatible agent can use it—Claude, ChatGPT, custom enterprise agents, and whatever comes next.

ATTOM's move was significant because it validated a model: property data companies can serve AI agents directly, not just human users. The question is: where does Australia stand?

Three Layers of Property Intelligence

To understand the Australian landscape, it helps to think in three layers:

These layers are complementary, not competitive. A property professional making a development decision typically needs all three: what's the site worth (Layer 1), what's the development potential (Layer 2), and what are the specific compliance requirements (Layer 3).

The API Landscape

Most major Australian property data companies have APIs, but the maturity varies significantly:

One company stands out: Dynamic Methods (Forms Live), which launched an MCP server in May 2025 for its real estate legal forms platform. It's the only Australian proptech company with an operational MCP server. But it serves legal forms and contracts, not property or planning data.

For property data—valuations, planning, compliance—no Australian company has an MCP server yet. Compare this to the US, where ATTOM already has 55 MCP tools in production.

The Unofficial Layer: Scrapers

Several unofficial MCP servers exist that scrape Australian property data from public websites:

• Domain property search scrapers (via Apify)
• Auction data scrapers
• Australian postcode lookup servers

These are brittle, unsupported, and lack provenance. They break when the underlying websites change their HTML structure. They have no SLAs, no data accuracy guarantees, and no accountability. For professional use—especially anything involving planning compliance or regulatory requirements—scraper-based tools carry significant risk.

Where the Gaps Are

The Australian landscape has clear gaps across all three layers:

Layer 1 (Valuations): CoreLogic and PropTrack have the data, but neither has built for AI agent consumption. CoreLogic's developer portal suggests API-first thinking, but the enterprise procurement cycle at these companies is measured in years. When they eventually build MCP servers, they'll serve the valuations layer—complementary to, not competitive with, planning and compliance data.

Layer 2 (Planning): Landchecker has zoning and overlay data via its Data on Demand API. Archistar has the broadest coverage with zoning, permitted use, FSR, and building height data, plus an “AI PreCheck” compliance feature. Neither has an MCP server. Archistar is the most likely to build one, given their existing API and AI focus.

Layer 3 (Compliance): This is the most underserved layer. Deep compliance checking—specific DCP provisions with clause-level citations, heritage provision chains, precinct-specific controls—requires structured extraction from source planning documents. No Australian company currently offers this at the provision level with full source provenance via an API or MCP interface.

Generative AI vs Deterministic Data

A critical technical distinction separates the companies in this space:

Some companies (notably Archistar with its AI PreCheck feature) use generative AI for compliance assessment. This means the AI reads planning documents and generates compliance opinions. The advantage is speed and breadth. The risk is hallucination—the AI can fabricate provisions, misinterpret cross-references, or miss exceptions that a structured database would catch.

Other companies use deterministic data retrieval—pre-parsing planning documents into structured databases, then performing database lookups that return verbatim provisions with source citations. No AI generates or modifies the output. The advantage is reliability and auditability. The limitation is that it requires upfront extraction work for each planning instrument.

For AI agents operating in planning workflows, the deterministic approach is more useful. Agents need reliable data sources they can cite. A provision returned with “Source: Inner West DCP 2020, Section 4.3.2(b), page 47” is actionable. A provision generated by an AI model that may or may not correspond to an actual clause is not.

Government Readiness

NSW Government has established the AI Assessment Framework (AIAF), mandatory for all government agencies procuring AI technologies. The framework requires assessment across five principles: community benefit, fairness, privacy and security, transparency, and accountability.

For planning technology companies, this creates both a hurdle and an opportunity. Companies that can demonstrate AIAF alignment—particularly transparency (source citations) and accountability (audit trails)—have a procurement advantage over those that cannot explain how their AI reaches its outputs.

The NSW Government's $2.7 million in AI grants to 16 councils signals active investment in AI-assisted planning. These councils have budget allocated and a mandate to adopt AI tools. They also have procurement requirements that demand the kind of transparency and auditability that deterministic systems provide by design.

What 2026 Looks Like

The consensus across industry analysis is clear: 2026 is the year of building agentic infrastructure for real estate and planning. The companies that build MCP servers and API-first data services now will be the default data sources that AI agents discover and use.

For Australian property technology, the opportunity is significant. The US has ATTOM. Australia has Dynamic Methods (for forms). For property data, planning, and compliance—the field is open.

The question isn't whether AI agents will consume Australian property and planning data. It's which companies will be ready when they do.