Heimstaden applies OpenAI tools to residential property operations
European residential real estate operator Heimstaden is expanding its use of OpenAI tools to change how it lets and manages homes. The company says virtual property-image production costs in the Netherlands fell from about €50 to less than €0.10 per image, a reduction of more than 99%, while Codex supported an environmental assessment of 20,112 property assets.
Heimstaden selected OpenAI in June 2025 after evaluating several platforms. Its stated aims are to equip employees with AI tools and redesign the processes at the centre of the business, with a longer-term ambition for buildings supported by increasingly self-operating systems.
AI applied across the letting workflow
When a home becomes vacant, teams must determine rental value, arrange imagery, create advertising and react to demand. Heimstaden is applying AI across those connected stages rather than treating them as isolated tasks. Tools built in ChatGPT Sites provide pricing intelligence and recommendations; AI supports virtual image styling and apartment descriptions; and an agent reviews live listings against the company’s quality standards.
AI-generated market feedback is intended to help letting teams make corrective changes and make pricing increasingly dynamic. Employees remain responsible for reviewing generated content, approving decisions and maintaining the required standards. Heimstaden estimates that the lower image-production cost alone could save €130,000 annually in the Netherlands.
Nikolaj Brammer, Heimstaden’s Chief Digital Officer, described the programme as resting on employee training and access to tools, a central transformation office and AI-champion network, and property-management systems and data models prepared for AI. The company’s approach to letting combines all three: local users work with the tools, the central team coordinates changes across the workflow, and property data is made accessible to pricing and listing applications.
Codex turns a portfolio analysis into a repeatable capability
Heimstaden also used Codex for an environmental-risk project spanning data held in public sources, different formats and datasets. Julia Erlandsen, an environmental sustainability analyst without a software development background, used the tool to support data discovery, cleaning, standardisation, quality checks and repeatable processing pipelines. She retained responsibility for the methodology and validation.
The work assessed 20,112 assets and generated more than six million rows of usable data. Heimstaden estimates the delivery requirement fell from five people working for a year to two people working for two months, and says it avoided the need for external consultants. The resulting capability is intended to support climate-risk assessments, reporting requirements and investor decision-making.
Adoption, controls and the next operating model
After ChatGPT Enterprise launched at Heimstaden in September 2025, half of employees were using it within two to three months. A central AI transformation office maintains a library of use cases, while approved tools operate under the company’s data-security rules. Staff are expected to review AI-generated material before sending it and to approve decisions themselves.
The company is also rolling out a Procurement Intelligence Platform that combines workflow automation with OpenAI models accessed through APIs. Designed to work across Outlook, SharePoint and Teams, it supports contract analysis, negotiation preparation and contract lifecycle management, with commercial decisions and approvals remaining with employees.
In parallel, Heimstaden is co-developing Amplio Estate, a property-management system intended to support AI-powered workflows through MCP. The practical implication for businesses is to start with a complete, valuable process, make the underlying data usable, and preserve clear human accountability as AI is introduced across systems.