TK Elevator gives service technicians AI agents built on Microsoft Foundry

From handwritten notes to a briefing on the phone

Microsoft describes the change through Torben Teichmann, a service technician for TKE. In the past, he might have had to ask around to find the elevator, track down a notebook with handwritten service notes and search through hundreds of pages of technical manuals. He might also have found he needed other tools, which meant trips back to his car. Notes were usually written up in the car, which he calls not very comfortable.

Today an AI app briefs him through his phone or tablet before he enters the building. He knows which tools to take, which elevator on which floor is affected, and what the issue is, whether regular maintenance or a service call for a problematic door. He can also search digital manuals and the elevator's previous service history on his phone.

After the job, a second AI voice agent asks about the work he performed. He answers in his own words, without writing or typing, and the agent asks follow-up questions if more detail is needed. His notes then become part of the record for colleagues who may service the equipment later. "I'm the boss at the job site and the AI is my buddy that gives advice," Teichmann says, according to Microsoft.

A decade of connected data behind the agents

The agents draw on a digital foundation TKE has built over the past decade by connecting elevators and escalators to the secure cloud. Connected-equipment data, service information and technical documentation are brought together using Microsoft Azure. TKE began with sensors and digital controllers and built a central data platform with Azure, Azure IoT and Azure Databricks, further developed in partnership with Microsoft Industry Solutions Engineering, now called Microsoft Frontier Company.

A common data environment mattered because TKE added different types of equipment through acquisitions. That meant servicing units made by other manufacturers as well as introducing its digital-native EOX and HELIX elevators.

Generative AI added a new kind of source. "Everything we did before was on structured data," says Christian Jung, head of digital data and AI foundation at TK Elevator. "But we were not able to leverage data from unstructured sources." Technical documents and technicians' service notes can now be extracted, searched and connected, which lets TKE give technicians technical guidance on how to fix a problem, according to Jung.

Digital operations centers and the scale of the work

TKE has created nine digital operations centers (DOCs), where data analysts and technical specialists look for patterns that can help technicians spot emerging problems or prioritize maintenance. The DOCs track the status of connected units, recorded events and error codes, and use machine learning to calculate "health scores" that help prioritize maintenance work.

One example is an elevator door that opens and closes several times before the lift moves, then seems to work normally for days. Tobias Lüsebrink, who heads digital transformation for TKE Germany, says the DOC "will see this kind of abnormal operation" even though it is not a breakdown. The DOC can flag the behavior and share details such as which floor has the affected door. The technician then inspects the equipment and uses training and judgment to decide what to do.

Information flows back too. Debriefing notes help TKE refine future troubleshooting and identify new patterns.

The scale is large. TKE, based in Düsseldorf, has 25,000 service technicians who care for 1.4 million units globally, and says its units carry 1.5 billion passengers a day. Christian Fongern, CEO of TKE Germany, says: "Uptime is our mission critical statement here, and service is the heart of our business."

What it means for technicians and service work

According to the source, the agents are intended to make both planned and unplanned visits easier by helping technicians find relevant information more quickly. Because TKE has 25,000 technicians worldwide, learnings from other parts of the world can reach a technician like Teichmann, and his own notes can in turn help colleagues in Germany and elsewhere.

The article presents the AI as an aid rather than a replacement: the technician remains the one who inspects the unit and decides what action is needed.