How did an automotive leader bridge the digital divide with AI voicebots without sacrificing service?

The goal

To use voicebots as an alternative to online self-service to decrease the volume of low-value live contacts without diluting CX performance or reducing the levels of service customers expect.

The outcome

5%

immediate reduction in live contact volumes

14%

automation of CX services within 12 months

80%

CSAT

71%

containment rate

27%

drop in AHT

The challenge

As the financial services division of one of the world’s largest and most respected automotive brands — covering vehicle leasing, financing, insurance and mobility — the client has always prioritized removing friction and delivering value in the vehicle ownership experience.

The client’s long-term goal is to transform its operations and generate measurable value through the use of digital tools, such as AI and automation. However, it takes a cautious approach — driven partly by rapid hype outpacing proven GenAI use cases, and partly by a customer base averaging over 50 years old — as quick transformation risks alienating rather than aligning with their needs and expectations.

At the outset of the partnership, the focus was on delivering traditional CX services but quickly included uncovering agent tasks ideal for automation. Automation promised not only to reduce live contact volumes and more strategic deployment of team members but also the removal of unnecessary customer touchpoints, ultimately improving the overall CX. However, since customers overwhelmingly favor phone contact — even for basic actions like password resets — traditional intelligent automation (self-service, portals, chatbots) would not meaningfully lower live contact volumes, improve average handle time or unlock efficiencies.

The solution

Support delivered from Greece and Turkey to Germany

Industries
Banking & financial services, Manufacturing

Channels
Voice, email, chat, SMS

Language
German

To transform the business without alienating customers and provide the client with confidence in respect to AI adoption, genuine use cases, and return on investment, we decided to focus on voicebot technology and proposed a three-year, multi-stage project with clear objectives:

  • Automate 5% of CX services in the first year.
  • Scale to 20% automation in year two.
  • Expand further to 30% automation by year three.

To ensure consistency of service quality, we also agreed to meet these objectives while hitting the same CSAT scores in automated channels that our agents in live channels achieve (80%) and maintaining or improving the client’s existing 65% customer retention level.

We identified four initial use cases based on contact volumes and suitability for voicebot automation:

  • Password resets
  • Information and guidance on tax and exemption orders
  • Support for activating and using facial recognition for account access
  • Prequalification for specific financial products and services

These use cases led us to select two platforms: Cognigy and our proprietary LLM voicebot platform, the Foundever AI Agent Foundry.

Cognigy was chosen for its robust NLU capabilities, scalability and effectiveness in scenarios with clear step-by-step resolutions. For other use cases, we used our AI Agent Foundry for greater flexibility in building a tailored product roadmap and enabling rapid implementation — a bot can go live in just eight weeks and be refreshed every 14 days with updated data and prompts.

We also chose the Foundry for its conversational flexibility, which is essential when customer engagements deviate from predefined paths, allowing us to quickly adapt bots to emerging inquiries, boost automation, reduce calls and improve CSAT.

We adopted a staggered approach, launching the bots individually after pilot phases over nine months.

Results

5%

immediate reduction in live contact volumes

14%

automation of CX services within 12 months

80%

CSAT

71%

containment rate

27%

drop in AHT

Within three months of the first password reset bot going live, we saved our agents 10,000 minutes that would have otherwise been spent handling live queries and achieved the year-one 5% automation target. This increased to 10% following the launch of the tax and exemption orders solution. During this time, we maintained an average CSAT of 80%, performing comparably to human-assisted channels.

With additional voicebot rollouts, automation rose to 14% by mid-2025. To date, our AI agents have successfully handled over 120,000 customer queries. The average session lasted 4 minutes and 51 seconds, showing the bots’ ability to manage complex interactions and deliver genuine resolutions.

The impact is also reflected in AHT for live channels, which has declined month over month since the initial bot went live. In October 2024, AHT was 419 seconds, and by July 2025, it had dropped to 308.