TK Elevator (TKE), one of the world’s largest providers of elevators, escalators and vertical transportation systems, is helping cities and buildings move beyond traditional mobility. TKE set out to use data and AI for predictive maintenance and AI-assisted diagnostics for field technicians while delivering equipment insights to customers through a digital portal. However, fragmented systems and limited access to IoT data made it difficult to scale these initiatives. TKE adopted the Databricks Platform to unify and process more than 500M daily events across its global operations. TKE can now deliver predictive insights that are designed to improve elevator reliability, uptime, availability designed to help keep people, workers, businesses and life as we know it today moving.
Moving elevator maintenance from reactive to predictive service
Keeping people and businesses moving requires more than just elevators — it demands reliable, data-driven operations. TKE collects more than 500 million telemetry events per day from connected elevators worldwide. The team’s goal is to monitor elevator health and predict service needs, shifting from reactive fixes to proactive maintenance. By analyzing this data, TKE aims to detect abnormal patterns in elevator behavior and identify early signs of component wear and operational anomalies before they lead to service disruptions that could impact public mobility.
In parallel, TKE aimed to modernize its Digital Operations Center (DOC) by turning insights from MAX, its IoT and predictive maintenance platform, into actionable guidance for field technicians. For example, a technician might receive a recommendation such as “inspect the door on the 5th floor” or “check sensor alignment,” tailored to each elevator’s condition. This approach reduces repeated service calls and ensures technicians arrive on-site with the information needed to resolve issues efficiently. Over time, these insights are designed to help prevent systemic issues and improve overall service quality. TKE also sought to deliver value to customers through a self-service digital portal, providing access to operational status, service history and equipment information.
Managing IoT data at this scale poses challenges for any organization. Telemetry signals such as door cycles, error data are generated continuously, and traditional relational databases were not initially designed to handle this volume and velocity. At the same time, TKE needed to integrate telemetry with data from fragmented enterprise systems and unstructured sources such as manuals, service records and contracts.
Since these systems were distributed across regions and business units, gaining a unified view of operations and equipment performance was a huge struggle for us. Even as predictive insights began to emerge from MAX, we faced the challenge of operationalizing them, explained Christian Jung, Head of Global Digital, Data and AI Foundation at TKE.
TKE wanted a unified platform to centralize data for predictive maintenance while also providing technicians with actionable insights to resolve issues more efficiently, in real time.
Building a unified data and AI platform for elevator operations
TKE began their migration to the Databricks Platform. Delta Lake serves as the core storage architecture, consolidating elevator telemetry, service records and enterprise operational data into a unified platform. “By leveraging a medallion architecture, we structured our data into bronze, silver and gold layers. This framework makes it possible to ingest raw telemetry data from connected elevators, with a goal to standardize data across operational systems and curate datasets for analytics and AI workloads,” remarked Christian. Delta Sharing helps securely distribute organized datasets across operational systems, like Salesforce, external analytics partners and data science collaborators. Lakeflow is used to orchestrate data pipelines as data moves through the bronze, silver and gold layers. This is designed to provide engineers greater visibility into data lineage and with a goal for continued increased transparency as systems learn.
Unity Catalog provides role-based access controls and detailed lineage tracking, helps to protect sensitive operational and customer information while helping engineers understand how datasets are created and used. TKE also leverages Databricks’ AI tooling, starting with Databricks Assistant, which helps to accelerate development by enabling engineers to write code, debug , and explore data. This automation is designed to enable teams to focus on building predictive maintenance models and deeper analytics that directly support technicians in the field.
MLflow is designed to bring consistency to model experimentation and evaluation, designed to power predictive models that have the aim to detect potential elevator issues in advance. Tools like AI Gateway and Vector Search are designed to allow teams to pull insights from structured and unstructured data, while Model Serving helps to deploy trained models into workflows and applications that technicians could use daily.
Supporting these AI workloads, TKE utilizes Databricks’ serverless compute to scale analytics and processing across the platform.
Designed to increase service efficiency across connected elevators
Running MAX and DOC on Databricks, TK Elevator has designed a process to take volumes of IoT data and create actionable insights designed to improve service operations – ideally at scale. Today, TKE generates 30,000–40,000 DOC tickets annually, reducing technician troubleshooting time on call-backs and unplanned shutdowns by up to 20%. Data from connected elevators can potentially be used and are designed to offer better service than non-connected units. And, predictive maintenance can be used to potentially lower costly emergency dispatches while the customer portal can enable additional subscriptions from equipment analytics.
The platform has enabled TKE to scale from just three markets to over 40 markets, connecting elevators and escalators — a more than 10x expansion. Data pipelines that once took months can now be delivered in weeks, aimed to accelerate analytics and ML development across the business. Meanwhile, data sharing with internal and external consumers has dropped from three months to three days, a 90% reduction that equips technicians with insights sooner. Centralizing operations on a single data lake across 30+ countries has unified enterprise and IoT data into one governed foundation, enabling TKE to develop agentic workflows and deliver reporting at scale.
We’re exploring how agentic AI can further enhance customer satisfaction, elevator uptime and availability and how it could potentially be used to enhance our capabilities across our global service organization,—Christian Jung, Head of Global Digital, Data and AI Foundation at TKE
These capabilities keep the human-in-the-loop, augmenting technician expertise — addressing the growing complexity of an always-on world and the expectations of a new generation that demands intuitive, tool-supported workflows. At Hannover Messe, TKE showed how its foundation — from IoT and predictive maintenance to a unified data and semantic layer — is designed to enable agentic AI to combine technical, service, and commercial context in real time, guiding technicians and creating a continuous knowledge loop.
Over time, agents can collaborate across systems and organizations, designed to enable new ecosystems for smart buildings and connected cities. With Databricks as the foundation, TKE’ solution is designed to scale connected infrastructure, governed data, and AI into tangible operational value.