AI NewsGo-to-marketAnnouncement
Heimstaden says ChatGPT cut Dutch staged property images 99 percent
OpenAI says Heimstaden cut the cost of a staged Dutch rental image from about €50 to under €0.10. One environmental study that was quoted at five people for a year took two people two months.
Image: Unite.AI
Why it mattersA €50 to €0.10 drop per image is the kind of number a procurement team will quote back. The source is the vendor.
OpenAI published a new customer case study on 3 November 2026 with real numbers behind it. Every one of those numbers comes from the vendor. The customer is Heimstaden, a European residential landlord with about 160,000 homes across Europe. OpenAI says Heimstaden uses ChatGPT Enterprise and Codex across letting, property operations and environmental risk work.
The headline figure in the case study: a virtually staged property image in the Netherlands that cost about €50 now costs under €0.10. OpenAI says that is a drop of more than 99 percent per image. OpenAI estimates €130,000 in annual savings from that one market and that one task.
The environmental study
OpenAI's second number is on a one-off piece of climate-risk analysis. Heimstaden assessed 20,112 of its assets and generated more than six million rows of structured data. OpenAI's case study says the work was quoted at five people for a year by external consultants. In the end, two people did it in two months. Heimstaden skipped the consultants.
The adoption claim
OpenAI says ChatGPT Enterprise reached about half of Heimstaden's workforce within two to three months of its September 2025 launch. The company has around 160,000 homes and operates in about ten European countries. The adoption number is a share, not a seat count. OpenAI did not publish the exact seat count in the piece.
OpenAI also described a Norwegian caretaker. The caretaker used ChatGPT to walk through a neglected apartment and build a priority refurbishment plan. OpenAI did not publish a saving in that case. The example sits as colour alongside the two main numbers.
Where the numbers come from
Every figure here is OpenAI's statement about its own customer. OpenAI studied two datasets to publish the case: it names Heimstaden's own operational data for the Dutch imaging work and the environmental analysis. The piece links no independent audit of the €0.10 figure. A reader with a procurement budget should treat these as vendor claims. The customer has not repeated them in its own voice yet.
The case still matters for one reason. A property management company is not an AI-native customer. It is a traditional operator buying ChatGPT Enterprise seats and plugging Codex into a workflow its staff already runs. Numbers at this grain, costed per task, are what most buyers look for. The question they ask is whether an enterprise AI rollout will show up on the P&L. OpenAI gave them two: 99 percent on a per-image task, and five people-years to four people-months on a one-off audit.
Source
OpenAI, "Heimstaden is changing residential real estate management with AI". Secondary reporting: Unite.AI.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.
