What drove the decision to develop an AI-based use case?
The topic of e-mail communication has long played a major role at GRENKE AG as a digital company. We receive many customer inquiries by electronic mail, sent to dedicated central mail addresses. Until now, these e-mails have been forwarded manually by the subsidiaries to the responsible clerks. Due to a good business development and process changes, a backlog of approximately 25,000 emails had accumulated in the central mailbox in Q1/2019 in one of our large European country organizations. Due to a lack of time and personnel resources, this volume of customer inquiries could not be processed manually fast enough - especially as it continued to grow daily. For this reason, we looked into digital solutions with automated analysis, classification and distribution of the emails.

What were your expectations or requirements for the AI application?
The requirements for the AI application were in the area of "Conversational AI". It was about a semantic analysis of queries in six languages (German, English, Italian, French, Spanish and Portuguese). Particularly important were the possibility of classifying the recognized texts according to defined attributes, simple system operation (SAS or own data center), secure communication and simple interfaces for the subsequent robotics process automation platform UIPath (RPA). Customer queries needed to be qualified faster and solved in an automated way through the use of AI. What could not be solved had to be forwarded to the right team as quickly as possible.

Which partner did you rely on with which technology for the development?
We chose the Cogito system from Expert System Deutschland GmbH as our development partner because the first application scenario did not have to be developed for Germany and Expert System is globally positioned in text analysis using AI.

The use of AI solutions must also be accompanied by a change in competences. How have you dealt with this challenge in respect of your employees?
We were able to integrate the system into our RPA platform relatively straightforwardly. The challenges for new competencies were therefore limited. The training and setup of the AI was handled by the manufacturer in this first use case.

In what timeframe did they implement the AI solution in your company?
Q3/2019 – Q4/2019

A conclusion to your AI solution:
All in all, this first deployment of an AI tool at GRENKE worked well. The challenges were no different from those of a "normal" software implementation. The experience with an existing RPA platform for easy automated processing of the results helped us a lot to implement the project quickly. Automatic processing is at approx. 80%.

One final piece of advice/tip to other entrepreneurs looking to apply AI:
It doesn't always have to be the really big thing. At GRENKE, we also started small. We are gaining more experience bit by bit.

More info at:
https://grenke.de/