Stadtwerke Witten: How an AI phone agent resolves requests and relieves customer service

Stadtwerke Witten supplies its customers with electricity, gas and water and also operates the indoor swimming pools in Witten, Germany. The utility already works with LoyJoy for chat (read the success story), and since autumn 2026 also on the phone.

Stadtwerke Witten at a glance

• Industry: energy and water supply, swimming pools
• Region: Witten, Ruhr area, Germany
• Channel: central customer service hotline
• Solution: AI phone agent on the LoyJoy Platform

Key facts about the project

• Around 50 % resolution rate: requests are resolved directly during the call, compared with under 5 % with the previous phone bot.
• More than ten use cases, including monthly payment changes, meter readings, callback requests, tariff advice, payment plans and information on swimming pools, events and e-mobility.
• Speak freely instead of choosing from a menu: callers describe their request in their own words, even several in one call.
• Noticeable relief for the service team on the hotline.
• Available 24/7, including outside service hours.
• Sovereign operation: the phone AI runs in its own EU data zone.

From a rigid phone bot to an AI agent

Before LoyJoy, Stadtwerke Witten already used a phone bot on its hotline. Tim Henne from the private customers department describes its limits as follows:

• It could only take callers through fixed, predefined scenarios.
• Dynamic conversations and changing the topic were not possible.
• In the end, it was “a slightly better answering machine”.

The decisive moment for the switch was the first test with the LoyJoy AI agent. For the first time, the team could speak freely with the system and discuss practically any relevant topic. That made the goal clear: not just to take requests on the phone, but to resolve them.

How the project went

For Tim Henne, it was a “very smooth start”, and the AI agent “worked superbly” from the very beginning. The most important step was setting up the phone system correctly.

The real work lies in the details: defining which requests the agent handles end to end and which details it needs for them, and then refining it step by step. A few examples of how the agent is tailored to the utility’s processes today:

• Service hours: before every transfer, the agent checks whether the service team is available. Public holidays and closing days are stored.
• Monthly payments: the agent records payments both for individual utilities and as one combined payment for all utilities.
• Tariffs: the agent asks which tariff the caller is interested in instead of listing all of them, and prepares an offer by email if the caller wants one.
• Swimming pools: the agent asks which pool the caller means and then answers questions about prices, opening hours, discounts and more.

Stadtwerke Witten and LoyJoy worked through open points together on a regular basis during the project phase. Stadtwerke Witten describes the collaboration as very close and good, with good support before and after go-live. The fact that the LoyJoy team responded quickly to changes was very important and valuable to them.

What the AI agent handles today

At Stadtwerke Witten, the agent most frequently takes care of monthly payment changes, meter readings, callback requests, tariff advice and general questions about products and services. Beyond that, it handles many other tasks:

Every request is passed on in a structured form to the responsible department. Names and numbers can be spelled out, and the agent reads back amounts and meter readings for confirmation. This way, the details arrive correctly and rework becomes less frequent.

Results: relief for the team, added value for customers

After just a few weeks in live operation, the results are clear:

• Around 50 % of requests resolved: directly during the call, without the service team having to step in.
• Less pressure on the hotline: the service team can focus more on complex requests.
• Available around the clock: this is very important to the utility. Customers can clarify many topics at any time instead of having to stick to fixed opening hours.
• Very good acceptance overall: there is only occasional criticism, which the utility had expected. Customers are also getting better and better at talking to an AI.

“My conclusion after just a few weeks: a much more flexible customer hotline, less pressure and real added value for customers, who reach the solution they need faster thanks to AI.”

Tim Henne, Private Customers, Stadtwerke Witten

Advice for other utilities

Asked what he would tell colleagues at other municipal utilities, Tim Henne remains realistic:

“It takes time and a lot of work, but the switch can pay off if you know what you want to achieve for your customers and your service.”

Tim Henne, Private Customers, Stadtwerke Witten

In Witten, this meant clear goals for every request and the willingness to refine the agent together during live operation.

Outlook

In the long term, Stadtwerke Witten wants to automate more and more. To get there, the AI agent will grow step by step with additional use cases.