Intelligent automation for insurance: 700,000 emails processed efficiently

Our client is one of the leading insurance groups in CEE and understands the immense importance of first-class and efficient customer service, which must be both timely and of high quality, no matter how demanding the customer’s requests may be.

deepsearch has made it its mission to support companies in precisely these areas. Thanks to intelligent automation, the use of our AI deepassist not only leads to a reduction in workload and thus to cost savings, but also raises the quality standard of customer service to a new level.

Optimization by deepassist

over 700,000

AUTOMATICALLY CATEGORIZED E-MAILS

- 22%

average handling time

Kunde

OOWV

Industry

insurance

company size

25,000 - 50,000

The challenge lies in the requests

The sheer volume of customer inquiries, especially emails, that our customer receives is impressive: with over 1 million messages, including over 700,000 emails per year, which are categorized manually and then processed by a large team with dozens of agents, the workload is enormous. This manual categorization ties up valuable resources and consumes thousands of working hours per year, which could be reduced by intelligent automation. High staff turnover, which is not unusual in customer service, puts an additional strain on processes.

deepassist is...

Industry specific

The deepassist AI already has the relevant knowledge of the insurance sector and thus reduced the training effort many times over.

Strategic

The use of deepassist as a C-NLU (Causal Natural Language Understanding) solution can be used by all departments within our client’s enterprise.

Intelligent automation is the solution

For our customer, it was clear that a secure, future-proof solution was needed that would not only overcome this challenge, but also lay the foundation for substantial simplifications through intelligent automation in the coming years. It could not be one of the many silo solutions with high technical debt that are advertised in the current AI hype.

deepsearch understands these requirements and offers specialized, pre-trained language models that can be used for all communication channels, telephone, email, chat, etc., and in conjunction with all applications, telephone systems, ERP, CRM, etc. Step one was the intelligent automation of the analysis and categorization of incoming emails. Building on this, the next project phases will support agents during telephone calls in real time to provide the right answer more quickly.

Another critical aspect is the clear transparency and traceability of AI-supported decisions – especially in contact with customers. With deepassist, not only is GDPR compliance ensured, but every decision is also transparently comprehensible (Explainable AI).

AI relieves employees of repetitive tasks through intelligent automation, allowing more time for higher-quality processes and making it easier to train new employees. Understanding every customer request in a context-specific manner and extracting important information such as policy numbers, IBANs, etc. makes deepassist a strategic natural language understanding service.

Eine win-win-win Situation

Trotz Ressourcenknappheit konnten wir unsere Anforderungen dank der vortrainierten KI von deepassist kurzfristig umsetzen, unsere Mitarbeiter:innen deutlich entlasten und unseren Kund:innen eine verbesserte Erreichbarkeit bieten. Strategisch haben wir mit deepassist eine Grundlage für weitere KI-gestützte Anwendungsfälle im Kundenservice geschaffen.

The result with deepassist

The results speak for themselves: the intelligent automation of email categorization already exceeded the required recognition rates in the pilot phase. Thanks to the pre-trained language model provided by deepsearch out-of-the-box, the training effort was minimal. The significantly reduced manual effort is an advantage for customers and is already evident in the initial use case with a 22% reduction in average handling time (AHT) for emails.

At least as important is the foundation that has been laid for the next AI-supported use cases. In the future, real-time support for employees will reduce the amount of training required and further shorten the AHT.