A voicebot answers customer service calls using natural language, understands the inquiry, and resolves standard cases independently or hands them over to a human in an organized manner. For companies in Germany, two things matter: a telephone channel that is accessible without waiting times, and an architecture that combines data privacy with high voice quality. This guide outlines the use cases, the distinction from classic phone greetings, and the criteria for selecting a provider.

At a Glance

• Around the clock on the phone: Standard inquiries such as booking appointments, status information, or address changes run automatically, even outside of service hours.

• Natural language instead of a button menu: Callers state their inquiry in their own words, and the voicebot understands dialects and filler words too.

• Data processing in Germany, connected to existing telephone systems via SIP trunk.

• Structured handover to employees, as soon as a case requires context or a decision.

• Channel switching during the conversation: from the call to a message containing a link or document, without losing the thread.

• Measurable impact: in use, up to 70 percent fully resolved conversations and a call duration of around 45 seconds instead of several minutes in the old menu.

What a Voicebot Achieves in Customer Service

A voicebot is a voice assistant for the telephone that connects Conversational AI with the telephone system. Callers describe their inquiry freely, the system recognizes the intent, and answers in spoken language. Unlike a recording with fixed announcements, the voicebot conducts a conversation, asks follow-up questions, and works with knowledge retrieved from the company's systems.

The benefit is clearly visible in accessibility. A large portion of service calls concern recurring questions: Where is my delivery, how do I change an appointment, what documents do I need. A voicebot answers these cases immediately, without any queue. Employees gain time for inquiries that require experience and judgment.

Voicebot Instead of Classic IVR

Classic IVR works with a rigid menu and the prompt to press a digit. Anyone whose inquiry does not fit the predefined structure ends up in the wrong queue or hangs up. A voicebot replaces the button menu with an open conversation.

The difference shows in three areas. Callers speak freely instead of navigating through levels. The system recognizes the inquiry and routes it directly to the matching process. And the call duration decreases noticeably because the path to the answer is shorter. In practice, this results in calls of around 45 seconds, whereas a menu-guided call takes several minutes.

Telephone Use Cases

Voicebots are suitable for inquiries that occur frequently and are clearly structured:

• Status inquiries: requesting delivery, order, transaction, or application status.

• Appointments: booking, rescheduling, or canceling, synchronized back to the calendar system.

• Master data: changing address, bank details, or contact details, with verified verification.

• Initial qualification: capturing the inquiry, categorizing it, and routing it to the right place.

• Callback requests: recording and scheduling outside service hours.

For sectors with high call volumes such as retail, energy providers, banks, or public administration, the voicebot thus covers a large part of the first-level support. Anything requiring context or a decision is passed on to a human.

What to Look for When Choosing a Provider in Germany

The market for voicebots has grown, and the offerings differ significantly in five aspects.

Data Location. A voicebot processes call content, often containing personal data. For GDPR compliance, where this data resides is what counts. Processing in Germany without transmission to third countries is the robust approach. Ask about the hosting location and the data processing agreement (DPA).

Architecture Against Hallucinations. On the phone, an incorrect response is particularly noticeable because there is no visible source. It is crucial that answers originate from verified knowledge and are not freely invented by the language model. A system that grounds its answers in the company's knowledge base delivers reliable information. Read more in our article on hallucinations in AI chatbots.

Telephony Integration. The voicebot must dock to the existing telephone system. A SIP trunk to common systems such as Avaya or Genesys is the prerequisite for the bot to go live without rebuilding the infrastructure.

Voice Quality and Dialect. German callers speak with dialects, filler words, and self-corrections. A voicebot that understands this keeps the conversation flowing. Test the recognition with real voices from your customer base.

Compliance Proofs. For payment data over the phone, PCI-DSS is relevant. The EU AI Act requires that it remains recognizable that an AI system is speaking. A provider should be able to provide documented proof of these aspects.

How Mercury Voice Solves the Telephone Channel

Mercury Voice brings Conversational AI to the telephone while following the same architecture as the chatbot. An orchestra of specialized models recognizes the inquiry, verifies the appropriate source, and only formulates the answer at the very end. The language model verbalizes; it does not make decisions on facts. Thus, the response remains bound to the verified knowledge of the company.

The models used are licensed, self-hosted Mistral models. No API calls are made to external providers; the data remains in Germany. The voicebot connects via SIP trunk to existing systems such as Avaya or Genesys, supports payment processes according to PCI-DSS, and switches between phone call and text message during the conversation. Mercury currently offers the voicebot in German and English, with other languages available on request.

Mercury Voice is currently in its beta phase. In practice, the channel achieves up to 70 percent fully resolved conversations and a call duration of around 45 seconds. A detailed overview is available on the Mercury Voice page.

Frequently Asked Questions

What is the difference between a voicebot and IVR?
An IVR works with a fixed button menu and pre-recorded announcements. A voicebot understands spoken language, conducts a conversation, and routes the inquiry to the appropriate process without menu navigation.

Is a voicebot GDPR-compliant?
A voicebot processes personal data and is therefore subject to the GDPR. Operation becomes compliant through processing within the EU, a data processing agreement, and binding the answers to verified sources. With Mercury, the data is stored in Germany.

How does a voicebot connect to the telephone system?
Via a SIP trunk to common systems such as Avaya or Genesys. This allows the voicebot to go live without rebuilding the existing infrastructure.

Which languages does Mercury Voice support?
Currently German and English, with other languages available upon request. The chatbot covers up to 98 languages.

For which inquiries is a voicebot suitable?
For frequent, clearly structured cases such as status inquiries, appointments, master data changes, and the initial qualification of inquiries. The bot hands complex inquiries over to employees.

Conclusion

A voicebot makes the telephone channel accessible around the clock and relieves customer service of recurring calls. For companies in Germany, the combination of data location, reliable architecture, and a stable connection to the telephone system is decisive. Mercury Voice brings these elements together and follows the same source-bound architecture as the chatbot.

Would you like to automate your telephone channel? Contact us or take a look at Mercury Voice.