Not long ago chatbots were seen as the next hype, that they would deliver all of our online information and AI would automate every aspect of business. Then it all went a bit quiet. We felt it was time for an update so we spoke to Xaver Lehmann, co-founder and CEO of customer-service chatbot startup E-bot7, about the industry in 2018 and how young companies can benefit from the technology.
Welcome Xaver and thanks for talking to us. Can you give us your initial thoughts on the chatbot hype – what happened there?
Firstly, at the start people thought chatbots would work in every sector, which simply wasn’t the case. Expectations were far too high if you like. The truth is that there are many applications for bots that just don’t make any sense, where an app works so much better – my example is always news & publishing bots! Secondly, companies that did put a chatbot on their website didn’t really explain to their customers what it was for. So they had customers typing all sorts of general questions in like “How old is Angela Merkl?”and “How did Bayern Munich play today?”, even in bots that simply weren’t built to answer these questions. Thirdly, companies thought they could develop a chatbot and it would immediately be able to answer all questions every customer had, forgetting they actually had to teach the chatbot the answers. But the truth is chatbots were never going to answer every question correctly immediately. It doesn’t happen overnight!
Can you tell us, why should a young company use a chatbot?
Speaking purely about our product (an integrated AI-bot for customer service), the size of the company doesn’t really matter. More important is the number of support tickets. If a company has a large number of support tickets where similar questions are coming in time and time again, then of course a chatbot like ours makes a lot of sense.
Are there any young companies from specific sectors which in your view would benefit from chatbots?
Telecommunications companies can always profit from chatbots, as the questions which come in are usually very similar and they have relatively few products. For companies with 100,000 products in portfolio it makes less sense, because there are just so many different questions which come in. A general rule: the fewer, simpler products a company has, the more suitable it would be for our chatbot.
What are the other really exciting applications for chatbots in your opinion?
For young companies it will be about two things: bots which automate much of the everyday grunt work and personalization.
In the first category is of course our bot and there are also other exciting applications such as A.xi where personal assistant bots help you to agree a time and a date for a meeting. By simply putting personal assistant Amy in cc in your emails, she is connected to your calendar and then takes over the correspondence with your contact to find a time and date which suits. Little things like that I really see really adding value.
In the second category there are bots used for marketing personalization. A company could develop a chatbot for Facebook messenger which they can use to deliver personalized content to customers. The bot instantly recognizes the profile of the customer they are talking to and adjusts the content – for example, a special offer on a particular product - accordingly. The best thing is, the customers could then buy the product directly via the bot.
How do chatbots learn to answer questions?
I can certainly explain how our bots learn. Ours are partially automated, quite simply we teach our bots answers to specific questions and each time a support ticket comes in the bot produces a suggested answer which then is sent to a customer support agent to check. If the agent agrees with the chatbot’s response they will send it on, if not they will adjust the response. The unique characteristic of our bots is that through this adjustment the chatbot learns automatically for the future. Often bots have to be manually taught new responses.
Lots of chatbots have their own identity (Alexa, Tay, Cortana). What is the advantage of this?
We have done many tests with customers and the vast majority were in favor of their virtual assistants having distinct identities. But under no circumstances should you lie to customers and pretend it’s a live chat. Customers need to recognize they are talking to a virtual assistant and that as such their questions should be short and to the point so that a chatbot can recognize the content.
Should staff be worried about their jobs?
No! In our case, we have shown very clearly that bots can answer very standard questions very well, but this is work which no employee would want anyway. Staff should be there to take on the more complex cases. Chatbots should be seen as additional help to manage the work load.
What does the future look like for chatbots?
Reality has hit companies which once thought you could use a chatbot for almost anything and integrated a chatbot just because they wanted to board the bandwagon. Now you see companies increasingly asking themselves where a chatbot could be of real value in their business. The future is more discrete, value-add chatbots for companies.
E-bot7 have recently been awarded a six figure investment from Commerzbank, and won the most Innovative Technology Award at the CCW Customer Service Congress. In early March they were also selected by the Impact Accelerator to participate in their Impact Growth Programme, winning €100.000 accelerator funding.