Skip to content

Frequently Searched

Tap to search.

Direct Contact

info@maibornwolff.de
Blonde Frau sitzt am Tisch vor Mikrofon Header_News_Weigerstorfer_mobile

How We Replaced Two Legacy Systems with AI at Weigerstorfer

Estimated reading time: 2 Minuten

How We Replaced Two Legacy Systems with AI at Weigerstorfer

A platform for the office and the construction site

Our client—Weigerstorfer GmbH—plans, builds, and maintains heating and plumbing systems, from the bidding process all the way through to the maintenance appointment. For a long time, this was handled using two separate systems: one for the office and an app for the technicians. Both had evolved over nearly ten years and were tailored precisely to the company’s workflows.

Our task: to replace both and consolidate everything into a single platform. Today, this platform handles all day-to-day operations: calculating quotes according to the GAEB industry standard, managing construction projects and procurement, and organizing maintenance, service, and warranty work. The technicians have their own iPad app and perform maintenance on-site entirely paperlessly. And then there was the data: over 250,000 orders with a project volume exceeding one billion euros that had to be migrated to the new system.

In this video podcast, CTO Alex Hofmann and Project Manager Stefanie Botz offer a behind-the-scenes look at the project:

A single agent for the entire team

At the beginning, the project didn’t even have an official AI. That changed quickly. Soon, the entire team—22 people at its peak—was working with a shared agent that knew the project wiki and every single ticket. That was a good thing, because we couldn’t access the code for the legacy systems. So our requirements engineering team had to figure out from the outside how the old systems worked, together with the business unit at Weigerstorfer.

Header_Weigerstorfer_2_klein
Weigerstorfer GmbH: Project Management Platform for Building Services Engineering

Things got really exciting during the data migration. With one of the two systems, there was simply no technical way to access the data. So Weigerstorfer hired student workers who transferred everything by hand: the old window on the left, the new one on the right. In the end, there were 13 of them. After a few weeks, however, it was clear that the work would never be finished.

Our idea: What a person can read from a screen, an AI agent can do too. The agent extracted the values from the interface, verified them, and transferred them cleanly into the new system. Where the old data was incorrect, we worked with Weigerstorfer to determine the correct values and used that information to create rules for the next run. In the end, well over 80 percent of the data was processed automatically.

Portrait von Stefanie Botz, Head of Affordable Software Factory bei MaibornWolff
Without artificial intelligence, we wouldn't have been able to finish it in time
Stefanie Botz, Project Manager at MaibornWolff

A nice side effect: The database is now neatly organized. This clears the way for the next AI ideas, such as in scheduling, quotes, or reports. The first experiments are already underway.