Have a Question About Chatbots? We have the answers.
Hundreds of interested technologists, marketers, customer service professionals and AI enthusiasts joined Tallan and Pluralsight for an informative webinar – “Chatbot Technology: The Innovative Enterprise Solution, hosted by Tallan’s very own Matt Kruczek. During that webinar one thing became very clear – chatbots are hot! Beyond the buzzwords, everyone wants to know how they can implement chatbots to benefit their business, and they came prepared with a number of questions. So we narrowed down the list to the most common, burning questions and answered them here for you!
With more people using Chatbots to make purchases, are you seeing (or anticipating) a backlash in consumer trust with the risk of hacks or security breaches?
There is always going to be an element of apprehension when it comes to utilizing any type of new service involving online payments. However, from a technology standpoint it’s important to note that most of the processes involved in payments over chat are not using anything new. In most cases, you will be connecting to your payment method via a web form that is only “initiated” by chat. Therefore, it’s really no different than the online payment methods that people are already used to.
What is the difference between a Chatbot and a personal assistant? Or are they the same?
To answer this question, you should understand a little bit more about the Chatbot landscape. It breaks down into three parts. The first part is the channel. A Chatbot channel is something like SMS or Facebook Messenger. It’s the place that users go to interact with your Chatbot.
The second part is the Chatbot itself. These Chatbots are specific instances that have a particular purpose. For example, the 1-800-Flowers Chatbot is a Chatbot that’s sole purpose is to service 1-800-FLOWERS customer’s needs.
Finally, the third part is something that I refer to as a “super” Chatbot. A super Chatbot combines the attributes of a channel and a Chatbot into one item. Take Alexa for example. Alexa has a lot of general Chatbot functionality. You can ask it things like “what time is it?” and “what is the weather going to be like today.” However, at the same time Alexa functions as a channel. You can create Alexa skills which are Chatbots themselves that sit inside the Alexa framework. So, you can ask things like “Alexa, ask Twitter for my Mentions.” In this case, you are accessing the Twitter Chatbot through the Alexa super Chatbot.
So to answer the question, I would label a “personal assistant” Chatbot as more of a “super Chatbot”. I.E. A Chatbot that can access many smaller Chatbots in order to function.
Assuming we wanted to ultimately publish the functionality to consumers of our products, but didn’t want to use “Hey Cortana”, “Hey Alexa,” “Hey Siri,” etc. — where would we go to develop our own, proprietary systems?
If I were architecting something like this I would create a codebase that would be very modular in design, making sure that I could add and remove functionality without a lot of overhead. Ultimately, there is no template for this type of thing; however, the tools to create it are readily available.
What basic things need to be considered, if we are going to develop a Chatbot based on machine learning?
The first thing that you need to consider when creating this kind of bot is how dynamic it’s going to have to be. Ultimately most Chatbots have a script that they are going to follow. It breaks down into “Intents” and “Actions”. If I’m creating a restaurant bot my Intents might be “Order Food,” “Get Hours of Operation” and “Make Reservation.” The “AI” ability in Chatbots is their way of being able to figure out what the users intent is without having to resort to keywords. In that way, you could say something like “Can I place an order” or “I want to get some food” and have the Chatbot be able to connect both of those phrases to the “Order Food” Intent. Once it knows your Intent it can respond with something like “Great! What can I get for you.”
However, if you start working in more Machine Learning, your script gets less structured. Now once the Chatbot figures out that the user wants to order food, maybe it uses Machine Learning to guess at what the customer wants based on past orders. In that case, its response to the user is going to be much more fluid, which is something your Chatbot will need to be able to handle.
My suggestion when trying to meld these two technologies is to pick specific touchpoints in your Chatbot script to utilize Machine Learning and to execute those points well. Don’t try to set the bar so high on your first go around that it ends in failure. Creating a Chatbot at that level takes a lot more time and effort than most people realize.
If we are developing Chatbot using Python, then which framework should be considered and why?
So, I’ll give two answers to this question, one is a direct answer and one is my personal advice. Right now, it is very possible to create a Chatbot utilizing Python. If this is the route that you choose then the way that I would tackle it would be to try and utilize all the Microsoft Cognitive Services API’s as I could. The reason being that you can call any of these API’s to execute specific functionality that is critical to building a bot without having to use a specific “framework” to do so. Stuff like LUIS and the Microsoft Bot Connector can be used purely through an API call.
However, while Microsoft does expose out a lot of functionality through API’s there is a lot to be said about using their frameworks. Trust me when I tell you that you don’t want to have to roll your own framework just to be able to write it in Python. My advice would be to learn either Node.JS or C# and use Microsoft’s. That way you don’t have to worry about both creating a Chatbot and creating a framework.
Is there any scope currently to connect the bot on WhatsApp?
The reason why there hasn’t been any bots built on the WhatsApp channel is because they don’t publically expose out any API’s. As soon as they make those available to developers, we should be able to create Chatbots for that channel.
What are some specific methodologies of implementation?
Microsoft has published a whole series of articles on this topic. You can read about it here.
To learn more about how Chatbots can boosts customer interaction and influence online sales over a myriad of channels such as Facebook Messenger, SMS, Slack, Alexa and much more, CLICK HERE.