At eHealth, our call centers handle on average 5,000 calls a day ranging from health insurance enrollment calls to customer service calls. These calls have a wealth of customer information which can be leveraged to improve customer satisfaction and call center efficiency. This talk will focus on some quick initial insights we obtained during our proof-of-concept phase for Natural Language Processing (NLP) using techniques such as sentiment distributions and topic models. I will also shed some light on the challenges we faced doing machine transcription with AWS Transcribe and how to overcome them. If you’re interested in NLP in general or specifically thinking about implementing NLP solutions in your call center, this talk should give you some ideas to get started.