How MEG automates 42% of customer calls with AI assistant Sophie

Mülheimer Entsorgungsgesellschaft mbH (MEG) is a municipal waste management service provider serving a catchment area of over 170,000 residents. Its four-person citizen services team handles hundreds of inquiries daily regarding waste disposal, bulky waste, recyclables, and bin management. 

Challenge

Despite a website relaunch, new app, and drop-off facility, MEG’s support call volume remained high. The lean service team was consistently inundated with recurring questions about opening hours, trash collection, and bulky waste.

“The demands on our waste advisory services are constantly increasing — whether due to growing public inquiries, the ‘Mülheim Cleans Up’ campaign with record annual participation numbers, or our facility tours.”

Timo Juchem, Managing Director of MEG

Externally, residents were being put on hold during peak hours and unable to reach any sort of support outside of office hours. Internally, the services team wanted more time dedicated to strategic projects such as waste management consulting and community engagement. MEG’s team looked to answer the question, “How can simple inquiries be answered more quickly, around the clock, without additional staff? The answer brought them to Parloa. 

Solution

On July 2, 2025, “Sophie,” MEG’s AI phone assistant powered on Parloa’s AI Agent Management Platform (AMP), went live. Implemented by service partner logen.ai, Sophie was trained on extensive FAQ knowledge regarding bulky waste, trash collection, container orders, hours of operation, and fees. Sophie greets callers with, “Welcome to MEG, I’m Sophie your digital assistant,” and continues to respond in natural language without involving a staff member. 

For more complex issues or uncertainties, Sophie routes citizens to the appropriate department, passing along all relevant information in compliance with European and German regulations on AI and data protection, so citizens don’t have to repeat what they’ve already shared in their conversations.

Rapid implementation through collaborative work

Close collaboration between MEG, Parloa, and logen.ai expedited the implementation process with specific project leads, short feedback loops, and rapid adjustments:

Test phase with daily updates

During the test phase, logen.ai iterated on Sophie daily. The MEG team tested Sophie in real-world scenarios and provided direct feedback to logen.ai, who adjusted the agent within 24 hours.

Post go-live support 

The day Sophie went live, the logen.ai team monitored all dialogs one-on-one during the first few hours and implemented hotfixes (immediate technical corrections). The hands-on support during this phase ensured that Sophie ran smoothly from the start and that any initial problems were resolved quickly and without downstream effects.

Subject-specific expertise

Sophie needed to be trained on specific waste management details, from trash collection schedules to recycling categories and municipal fee structures. Multiple rounds of iteration and careful curation of opening questions ensured Sophie could provide precise answers both technically and contextually.

Results

Within the first seven months post-launch, Sophie delivered results well above the industry average for conversational AI systems:

Operational excellence

• ~25,000 conversations handled without additional staffing costs and without capacity limits.

• 42.5% automation rate: Nearly one in two calls are resolved entirely by Sophie, exceeding the industry average of 30-40% .

• Handling time: Call resolution averages 1 minute and 12 seconds, significantly lower than the contact center standard of 4-6 minutes per call. 

• Immediate availability: Sophie answers every call right away, even outside of office hours, on weekends, and on holidays.

Service quality

• 88.1% engaged conversations ~20% more callers tend to stay on the line with Sophie than with other automated experiences.

• 53.8% FAQ usage:  Sophie leverages MEG’s  pre-existing knowledge base in over half of all calls. 

• 26.8% drop-off rate: ~15% less callers hang up on Sophie than on other conversational AI platforms. 

By automating routine calls, MEG’s four-person citizen services team can now focus on more complex cases and strategic initiatives.

“With Sophie, we’re creating new opportunities for projects related to waste management consulting – a real win for our team and the city.”

Thorsten Gerbitz, Department Head, MEG

Conclusion

By leveraging Parloa’s voice AI, MEG solved a common service problem: small teams stretched thin by fielding hundreds of calls a day, with residents spending too much time on hold. Seven months in, Sophie has already automated thousands of support requests and saved the four-person team hours of answering routine questions. Now, routine call volume no longer crowds out complex cases, and residents get the answers they need when they need them.