3 Questions for ... Stefan Riesel
1. Stefan, after 20 years at CreaLog, what still excites you about your job?
I'm still surprised myself by the range of industries and topics I deal with day to day, and by how different the solutions are that we build for our customers. In the process, I not only meet many interesting people but also gain deep insight into a wide variety of industries and the specific challenges they face in customer service. One day I might be discussing last-minute bookings via a voice portal for a major pay-TV broadcaster, and the next I'm talking to a furniture retailer that wants our VoiceBot to answer questions about product availability at the nearest store or delivery status. A week later, I'm sitting down with a public authority in Berlin that wants our VoiceBot to support the unified D115 phone number. Or with an energy provider that wants to capture meter readings via a VoiceBot and automate outage announcements.
In automation projects for customer service, my role as a consultant is called on almost every time, because our clients rely on our years of experience in this field. Artificial intelligence is the best example: if you believe the AI hype, all you need to do for customer service to work is install a chatbot or VoiceBot and the AI will answer every question on its own. Those are, of course, pure marketing claims, and it's my job to set my clients straight, especially when they're convinced that's exactly how it works and no other way.
The fact is that an AI first has to be trained intensively before it can answer customer inquiries satisfactorily. To do that, as many real customer conversations as possible should be recorded and analyzed. Our transcription solution, also known as speech-to-text, and AI-supported text analysis are a great help here. Even so, a good deal of manual work is still needed to correctly identify what customers actually want. Only then does the VoiceBot not just recognize the spoken word, but interpret it correctly and truly "understand" the caller's specific request.
2. You just mentioned chatbots and VoiceBots. Aren't those different technologies?
Only at first glance. It's true that a VoiceBot deals with natural speech, while chatbots, messaging services, or emails deal with the written word, but in the end all of these communications are processed as text, not audio. Through speech recognition and natural language processing (NLP), a phone conversation is also turned into text that can be responded to just like a chat.
That's why, right from the start of a project, we encourage our clients to consolidate their input streams from the various communication channels. That way, all the data for customer service, meaning intent recognition and the response database, comes from a single central source. It's not acceptable for a customer to get different answers from the VoiceBot, chatbot, or messenger, or to have the downstream processes differ from one another. Integrating our solutions into existing systems and connecting them to backend systems and databases is therefore a central part of the work our project managers and UI/UX designers do.
3. Do you develop core technologies like speech recognition and natural language processing yourselves?
For these technologies, we take a best-of-breed approach, using market-leading technologies from Nuance/Microsoft, Google, and Amazon. Together with our clients, we decide which technology is the best fit. Naturally, we also factor in data protection requirements and how the solution will be used, whether from the cloud or on-premise. When it comes to billing models, customer-friendliness and flexibility come first for us, which is why we offer a wide range of purchase, rental, and pay-per-use models for our complete solutions.
Stefan Riesel recently gave a webinar with numerous customer examples as part of the 9th Customer Service Week and answered questions live. If you're interested, you can watch it here: https://www.youtube.com/watch?v=tj5oGcQadao
For an in-depth discussion, connect with Stefan Riesel on LinkedIn: https://www.linkedin.com/in/stefan-riesel/ or email stefan.riesel@crealog.com
