BRENNEN

Industries / Construction and property

Real estate

The real estate industry buys, sells, rents out, values and manages land and buildings, and keeps the records of who owns them.

1858

South Australia's Real Property Act, passed on 27 January 1858, made ownership of land pass by an entry in a government register instead of a chain of private deeds; the system, named Torrens title after Robert Richard Torrens, spread to New Zealand, the other Australian colonies, Malaysia and some US states.

sahistoryhub.history.sa.gov.au

40

Milestones achieved across 6 domains.

9

To watch for: announced, not yet confirmed outside.

12

Open deliverables. Nobody has shipped them.

In one minute

Where it started

Land was rented under written rules in Hammurabi's Babylon, nearly 4,000 years ago. The modern trade took shape in the 1800s. South Australia made ownership pass by an entry in a public register in 1858, London surveyors formed the profession that values and manages property in 1868, and US real estate boards were pooling their listings by the 1880s.

Where it stands

Savills, a property adviser, puts the value of the world's real estate at $393.3 trillion at the start of 2025, of which $286.9 trillion is homes. Listings, price estimates and rental ads have moved online, and rules made since 2023 now govern US agents' pay, rent-setting software and repair deadlines in English social housing. Much of the daily work still rests on documents read and photos judged by people.

What is open

The open jobs are mostly public tests and datasets that nobody has made: which listing photos were altered, what a lease says, whether an AI title search missed a lien, what caused the damp in a flat, and what a building's data points mean. Companies sell tools for several of these jobs but publish only their own accuracy figures, or none. AI models that read documents and photos, and web agents that search public records, put this work within reach of a student or a small team.

Domains

AchievedWatch forOpen

Other narrower industries

Short-term rentalsPropTechHome inspectionSelf-storageCoworkingRV parksParkingCommercial real estate brokerage

Figureheads

4 people

Advances to build with

1 research note

Open in real estate

3 of 12 open deliverables

Open · Listings and search

Altered photo test set

A public set of listing-style photos with known edits, and a detector scored against it. Photograph rooms and house fronts you have the right to use, then make the edits listings use: a blue sky swapped in, a greener lawn, power lines or clutter removed, and virtual staging, which adds furniture that is not there. Keep every original, so each edited photo has a known answer.

Built with what exists

A phone camera; image editors with AI fill and virtual-staging services to make the edits; the C2PA Content Credentials standard, which can record edits inside the image file; an AI model that reads images, such as Claude Fable 5.2 (see the Note "Three working devices"), as one baseline; Label Studio for labels; Hugging Face to host the set.

Credit · unclaimed

Open · Valuation

A valuation test run ahead

A public leaderboard that scores automated valuation models only on sales that had not been recorded when each model was frozen. Each month, every entrant's model is saved with a timestamp. When HM Land Registry publishes the next month's prices, the leaderboard runs each saved model on those sales and scores it. No model can have seen the answers.

Built with what exists

HM Land Registry's Price Paid Data, which covers sales for value in England and Wales, is updated monthly and is released under the Open Government Licence; Energy Performance Certificate data for floor areas; the Cook County Assessor's published valuation code as a model to adapt; Python; a scheduled job on GitHub Actions, whose commit history gives each saved model a public timestamp.

Credit · unclaimed

Open · Sales and title

Deed-to-map accuracy test

A public test of how well software turns the written boundary in a deed into a shape on a map. Many US deeds describe land by metes and bounds, a chain of compass bearings and distances from a starting point. Collect modern deeds from counties that post records online free, take reference outlines from recorded surveys where they exist and from the county's parcel map otherwise, and score each tool's shape.

Built with what exists

County recorder websites and county parcel map layers; an AI model that reads documents, such as Claude, to pull out each call, with a second checking pass (see the Note "Seven more flagged"); Python with the Shapely geometry library; QGIS to inspect the results.

Credit · unclaimed