Case Study

Automating a significant amount of multi-turn requests in collaboration with human customer service agents

Situation

ERGO, a leading European insurance company, wanted to reduce costs in their customer service operations and allow for 24/7 uptime to improve the customer experience.

Challenges

Basic chatbots and IVR systems have been rejected by users as they’re usually more frustrating than helpful. Most solutions in the market only allow for basic, FAQ-based bots that don’t connect to backends and cannot handle complex customer requests. As ERGO wanted to keep the high quality standards from their award-winning chat agents as well as high data privacy requirements, none of these solutions worked for them.

Solution

After an initial proof of concept, ERGO implemented a production deployment of the Rasa Platform, which has the potential to automate over 30% of customer requests as a first step towards full automation. The identification of skills was based on an extensive analysis of thousands of real customer service conversations. In a training phase of 12 weeks, the agents were able to improve the accuracy of the system.

Next Steps

Continued integration of fully automated skills into the live chat environment.