Case Studies / arvato Bertelsmann
TELECOMMUNICATION · CALL CENTER OPERATIONS
Replacing manual call notes with automated AI summarization
arvato’s call center agents were spending significant time manually writing up call summaries after every customer interaction. We built an automated speech-to-text and summarization pipeline that standardized the process and eliminated the manual effort entirely.
Client: arvato – part of the Bertelsmann group, a leading international service provider with call center operations across multiple markets.
KEY RESULTS
100%
Call summarization fully automated – zero manual note-writing
Polish
Custom AI models trained for Polish language processing
Cloud
Scalable cloud-based infrastructure for high-volume processing
Uniform
Standardized note format across all agents and teams
INDUSTRY
Telecommunication
USE CASE
Automated call summarization
AI APPROACH
NLP + custom ML models
SYSTEMS
Call recording, CRM
LANGUAGE
Polish
ENGAGEMENT
Proof of concept → production
The challenge
arvato’s call center handled a high volume of customer interactions daily. After every call, agents had to manually write up a summary – capturing the customer’s issue, what was discussed, and any follow-up actions. This was time-consuming, inconsistent, and created an information bottleneck within the organization.
Hundreds of hours of call data were generated each day, but extracting structured, useful information from that data was entirely manual. The quality of summaries varied from agent to agent, making it difficult to track trends, share context between teams, or use call data for operational improvements.
The core problem: agents spent a significant portion of their time on administrative note-writing instead of handling the next customer. And the resulting summaries were inconsistent enough to limit their value downstream.
What we built
We developed an end-to-end pipeline that took raw call recordings and produced structured, standardized summaries – without any manual intervention from agents.
Speech-to-text transcription. The first stage converted call audio into text. Working with spoken Polish introduced specific challenges: variations in audio quality, interruptions, colloquialisms, and limited out-of-the-box language support compared to English or German. We developed custom post-processing techniques to refine transcription accuracy and handle these edge cases reliably.
AI-powered summarization. Once transcribed, the system generated concise summaries based on arvato’s internal procedures and note-taking format. We trained custom ML models to extract the most relevant information – customer intent, issue category, actions taken, and next steps – and present it in a uniform structure, regardless of which agent handled the call.
Cloud-based processing infrastructure. The entire pipeline was deployed on cloud infrastructure to handle the volume of daily calls efficiently. This allowed the system to scale with demand – processing more calls during peak hours without degradation.
The results
BEFORE
Agents manually wrote call summaries after each interaction. Quality varied by person. Call data was underutilized for analysis or process improvement.
AFTER
Summaries generated automatically in a standardized format. Agents moved to the next call immediately. Call data became a structured, searchable asset.
Manual effort for call summarization was eliminated entirely. Every call produced a consistent, structured note – regardless of agent, call complexity, or time of day. This freed agents to focus on customer interactions rather than administrative tasks.
Beyond the immediate time savings, the standardized summaries opened up new possibilities for data analysis – identifying recurring issues, tracking resolution patterns, and improving internal processes based on actual call data rather than fragmented manual notes.
Technology used
More Case Studies
See how we’ve helped other companies
AUTOMOTIVE · LEADING LUXURY MANUFACTURER
Intelligent virtual assistant replacing manual planning queries across SAP and cloud systems
Product planners spent hours manually querying SAP BW and multiple data warehouses for every decision. We built a bilingual voice-and-text assistant that retrieves planning data on demand – no system expertise needed.
Hrs → Sec
Data retrieval
DE + EN
Voice & text
SAP BW
Integrated
MANUFACTURING · RADAWAY
Making email-based order processing reliable with LLMs
Staff were manually reading customer emails, identifying products, and entering orders by hand. We turned a promising AI prototype into a production system that handles it end to end, across languages, formats, and attachments.
-90%
Manual intervention
95%+
Match accuracy
LOGISTICS · FR. MEYER’S SOHN
Eliminating manual data extraction from thousands of daily shipping emails
Operations staff were manually reading German and English logistics emails to pull out routing and scheduling data, every single day. We built an AI pipeline that extracts, structures, and delivers the data automatically.
–80%
Manual effort
2 langs
DE & EN
On-prem
Deployed