OpenAI case study: Heimstaden rebuilds property management with ChatGPT Enterprise and Codex

On November 3, OpenAI published a customer case study on Heimstaden, a European residential real estate operator, describing how the company folded ChatGPT Enterprise and Codex into everyday property management. Rather than a concept demo, the post is built around checkable numbers: image production costs, assets assessed and employee adoption.

Heimstaden’s leadership decided in June 2025 to accelerate its AI bet, choosing OpenAI after evaluating several platforms. The stated goals were twofold: empower employees and rethink the processes at the heart of the business. “The purpose of our buildings hasn’t changed,” co-CEO Helge Krogsbøl says in the post. “But the way we operate them continues to change.”

Letting was the first process reworked. When an apartment falls vacant, pricing recommendations come from tools built in ChatGPT Sites, listing images and descriptions are AI-generated, an agent reviews live ads against the company’s quality standards, and market feedback is summarised to help teams adjust prices. Employees still review and approve decisions.

The cost effect is the clearest signal. In the Netherlands, producing a virtually staged property image has fallen from roughly €50 to under €0.10 — a reduction of more than 99% — and Heimstaden estimates annual savings potential of €130,000 in that market alone. The bigger point is speed: homes reach the market with far less manual coordination.

Codex was used by people outside engineering. Julia Erlandsen, an environmental sustainability analyst without a software background, used it to discover, clean and standardise scattered data and build repeatable pipelines while keeping control of methodology and validation. The result was 20,112 assets assessed and more than six million rows of useful data, with an estimated requirement of five people for a year reduced to two people for two months, and no external consultants.

Adoption figures are equally concrete. After ChatGPT Enterprise launched in September 2025, half of employees were using it within two to three months, supported by a central AI transformation office that maintains a library of use cases. In Norway, a previously sceptical caretaker used ChatGPT to assess a neglected apartment and draft a prioritised refurbishment plan.

The next phase moves from answering questions to taking action. A Procurement Intelligence Platform now being rolled out combines workflow automation with OpenAI models accessed through APIs across Outlook, SharePoint and Teams, supporting contract analysis, negotiation preparation and contract lifecycle management. Heimstaden is also co-developing Amplio Estate, a property management system built to let AI tools such as ChatGPT retrieve information and act in core systems through MCP, with human oversight.

The case is useful because it locates the bottleneck precisely: adoption and imagination, not technology. Heimstaden’s leadership lessons are about people, local champions and getting the process right first, which suggests enterprise AI value comes from redesigning workflows rather than from switching on a model.