“Hi, my name is Devon. How may I help you today?” That was one of my first interactions with a chatbot many years ago. Chatbots are designed to help people through a series of interactions with a computer. Devon was the “person” “working” for my network provider and was there to help me resolve a problem.
That early Devon was very frustrating to work with. Devon had me do all the things I already tried (restart the computer, unplug the cable modem for 15 minutes, and plug it back in, make sure the router was plugged in, etc., etc.). Devon did not give me a chance to say, “Hey, I already did that.” Well, technically, it gave me a chance because I could type anything I pleased, but Devon didn’t care. Devon had a script to follow. Once I completed the steps, and there was still a problem, Devon told me to call technical support where I had to rehash the whole issue again with a real person.
The growing chatbot industry
Until recently, chatbots were always relegated to a support page to help solve common questions or a sales page to get people’s information and put them in touch with an actual salesperson. With the advent of search engines and community sites, it is easier than ever to get answers to questions even with some haphazard search queries. That means the chatbot market might be hot, but there is still a lot of work to do so they are recognized as useful.
Chatbots are finding a new home, though. You may not think of it this way, but popular messaging and AI services like Alexa, certain Slack apps, and Siri are not much more than dumbed-down (but highly useful) chatbots. From a consumer perspective, they are quite popular. Ask Siri, “What’s the weather,” and you get a contextual response. Siri knows I’m asking the question in San Francisco and knows what time it is, so it responds with exactly what I need. Companies like Amazon and Apple can even license data to give you more information such as the score for last night’s game or the latest breaking news. Those small one-off communications (ask a question, get an answer) may not be called chatbots, but that is what they are.
Then you have chatbots to solve more consumer problems, like ordering a pizza (listen to the attached T-Suite Podcast for some thoughts on this topic). You can order a pizza by saying, “I want a pizza” and then you go on to answer questions about the size, toppings, sauce, and even the type of crust. That may seem useful, and maybe it is to some, but at some point just placing an order on a website or app is probably easier than typing or speaking answers to all those questions.
Then there are business-level chatbots. Imagine you are in the field and need to fix a piece of equipment but there are a few models, and you require specific information on how to perform your task. In that scenario, you could have a chatbot that will help you get to the particular type of maintenance you need to complete and walk you through the steps.
Of course, there are much bigger opportunities for chatbots, but we have to find our way out of manually creating workflows to solve one-off problems.
Secret sauce to a human-interface chatbot: Ontology
In today’s T-Suite Podcast, Alex Masycheff, CEO of Intuillion, talks about the evolution of chatbots and how they will be much more useful in the future. He believes the real answer to creating useful chatbots is not just having content (how to reset a modem), and context (where I am located). You also need ontology. With ontology, you create content that is smartly and intelligently written and tagged; you build a knowledge graph of that content, and couple that with machine learning to deliver rich user experiences.
We talk about several examples, but I will share a useful one here. Let’s say you tell a chatbot the paper in your printer is jammed. That chatbot might kick off a prebuilt workflow that forces you down the path of opening the printer and removing the jammed paper. However, a person’s assumptions could be wrong. Perhaps the paper isn’t jammed — perhaps there is no paper at all. Maybe it is a wireless printer that lost its Internet connection or maybe you are out of a particular color of ink. A well-built chatbot can go off-script and help resolve the problem by using a vast knowledge base of interrelated content topics to help you solve the problem.
Alex can be reached via email at [email protected].
Special thanks to Scott Abel, the Content Wrangler, for introducing me to Alex for this T-Suite Podcast.
