Agentic AI drives real-time Service Intelligence

Generali Austria, part of the Generali Group, serves 1.7 million customers each year. Customer interactions required significant manual effort, while valuable insights remained largely untapped. To address this, Generali introduced an Agentic voice agent that supports service teams during live calls with real-time transcription, translations, summaries, and contextual suggestions. The solution analyzes conversations as they happen, streamlining documentation and improving efficiency.

Generali_Logo

3,127

Piloted voice AI sessions

1,000

Contact Center reps supported

5–7x

More positive feedback

Hear from Generali firsthand

The challenge

Generali’s existing telephony and system landscape was not designed for real-time AI integration. There was no direct support for speech-to-text processing, limited connectivity between contact center systems and AI platforms, and no fully optimized audio setup to ensure consistent transcription quality.

The organization also operates in multiple markets with different systems and levels of technological maturity, which made standardized AI deployment difficult within existing infrastructure constraints.

Operationally, large volumes of customer interactions required efficient processing, while service teams spent significant time on manual documentation and categorization. Valuable insights within conversations, such as sales opportunities and recurring issues, remained largely unused.

Generali Story Image
Sandra Generali
“Understanding customer intent at scale is one of the most critical challenges in customer service. With NiCE Agentic, we can finally move from reactive handling to truly intelligent, efficient, and consistent interactions.”
Sandra Öffel
Project Manager Process Automatisation @ Generali Austria

The solution

Generali piloted a proof-of-concept designed around real service team workflows. A dedicated button in the existing call interface allows support specialists to add AI to live calls via conference. At the same time, a web application opens simultaneously to display real-time outputs such as conversation summaries and detected intents.

The solution combines speech-to-text for live transcription, LLMs for structured summaries and information extraction, and intent recognition models to classify requests. Additional conversational capabilities were tested for future voice interactions, and a retrieval-augmented approach was evaluated to connect internal and external knowledge sources.

The web application serves as the central interface, consolidating all insights in one place. Service teams review summaries, validate categories, and provide feedback on accuracy, creating a continuous improvement loop across transcription quality, prompt design, and usability. Through iterative refinement and close collaboration between business and technical teams, the AI agent is continuously improved.

Generali Story man

The impact

The pilot delivered immediate value within just 8 weeks. More than 3,127 voice AI sessions in a live environment provided a clear picture of its performance in real service scenarios and received positive feedback.

The quality of automated summaries was rated predominantly good to very good, particularly in clarity and structure, reducing manual documentation effort. Intent recognition accuracy improved by 9 percentage points.

Service teams showed strong engagement and acceptance. The solution was widely perceived as supportive, non-disruptive, while web app optimizations further reduced post-call effort and improved user experience.

The pilot confirmed that Agentic voice use cases are technically viable in live operations and established the foundation for broader rollout and scale, including integration into CRM systems and service processes.

Discussion White Icon

Real-time Conversation Analysis

AI White Icon

Continuous Learning & Optimization

Document White Icon

Automated Summaries & Categorization

Want the same success with NiCE Cognigy?

Use AI to deliver better service and reduce costs, while supporting customers and service teams. Around the clock, at scale.