As part of our recently launched eBook about Chatbots, Conversational UI and the Future of Online Interaction, we had the pleasure and the opportunity to interview Matt Schlicht, Founder and Editor of Chatbots Magazine. Read his interview below and find out the recommended platforms and frameworks for chatbot development, how chatbots can help companies generate revenue, whether the increased usage of Messenger bots could lead to a decrease in the usage of mobile apps and many more chatbot-related insights.
[ebook06]Learn about the history of chatbots, the most relevant messaging platforms, innovative use cases and best practices from international companies![/ebook06]
1.As the editor of Chatbots Magazine, you must have heard of hundreds of chatbots by now. Which one would you say represented the biggest surprise?
Matt Schlicht: This is a really tough question because even though there are thousands of chatbots by now, I have yet to find one that I actually really, really like, and that I’ve been delighted by. Everyone that’s building chatbots has all the tools in front of them for doing so, but they have not yet figured out how to create an incredible chatbot that I would want to brag about in this interview.
On the other hand, there is a chatbot that I really like. I don’t know where it will go, and I don’t know how big it will get, but I was surprised of how simple it was and how much I liked it. It’s called Streak Trivia, a Facebook bot that sends out a notification at 1PM to anyone who has ever chatted to it before. It does a multiplayer true-or-false trivia game. If 50 people are playing today, it sends a question out to all of them, and whoever gets the answer right gets hit the next question, while whoever gets it wrong gets knocked out of the game. It’s a competition to see who can get the most answers right, and the last one standing wins the game for the day.
“Everyone that’s building chatbots has all the tools in front of them for doing so, but they have not yet figured out how to create an incredible chatbot.”
Chatbots, much like websites or mobile apps, can provide value in a lot of different ways, so this would be a must for a great bot to surprise me. Other than that, it should be lightweight and feature interactions. Some of the ideas would be better as a website or a mobile app, but companies forcibly turned them into bots and provide a bad experience this way. The last thing is that bots should be delightful in some way, and should make people happy when using them. One of the biggest things that people need to be careful about, whether they are companies using a chatbot or a developer inventing a chatbot, is the idea of sending massive amounts of messages (deals, coupons) to all the potential customers all the time. That wouldn’t make chatbots any different from spam over email, and people would start ignoring them. On top of that, platforms like Facebook, Skype or Kik would start filtering out the messages people get from bots because so many of them would be sending spam. So spamming people would be the number one thing not to do when creating a chatbot.
I noticed that when developers first start building a bot, they get very excited with the potential that the bot can have machine learning behind it and can have natural language processing, and they spend a lot of time incorporating different APIs from 3rd-party services to power machine learning and NLP in their bot. The problem is that this technology isn’t quite there yet. I mean, if anybody has talked to Siri before on the Apple iPhone, Siri doesn’t even work that well, and that’s with a massive company behind it. The chances that you could create something that’s better than Siri are really-really slim. If you can, that’s awesome, you should totally do it, but most people can’t. One other recommendation I have, at least for the next couple of years, is that any bots that you’re creating probably shouldn’t be reliant on any machine learning or NLP, and you should just focus on making them very-very simple and lightweight, and providing value in a very straightforward way, and not relying on too much conversational interaction.
2. What if every company out there had its own chatbot, either on Facebook or on their own website? Could these bots start interacting with one another to facilitate business processes?
Matt Schlicht: Yeah, I think that it could happen. However, it’s unclear exactly how that would work because a lot of companies already have APIs. If one company has a chatbot and for some reason, it wanted to interact with another company, and it had the option to interact with that company’s chatbot or that company’s API, it would make more sense in most scenarios to interact with the API. It’s definitely possible that bots would start talking to each other, but even if that sounds super cool, if it turns out that it’s more efficient for a bot to interact with an API rather than with another bot, then that’s most likely what’s going to happen.
3. For companies who are at their first chatbot, what platform would you recommend? What is the best framework they could use if none of their employees has software development skills?
Matt Schlicht: In terms of what platform you should build your bot on when you’re first starting out, it really depends on where your customers are and where your audience is. If you’re targeting a US-based audience, then Facebook Messenger is probably the best option, whereas if you’re based in China, you would probably want to build your bot on top of WhatsApp. That’s really how you would come to the decision which platform you should be building on. Based on where your audience is, where your customers are, what messaging platforms they’re using, you can select the platform where it makes the most sense to build your bot.
In terms of products that people can use to build bots without any programming experience, there are multiple companies who are trying to do this, but none of them is really that great right now. I’ve seen a lot of people who are trying to create bots on any of these platforms that promise that you can create a bot without programming, and everybody gives up. They spend four or five hours, and unless they’re very-very technically oriented people, they are not successful and get frustrated. So, a product that we can recommend to any business out there and that can be used to create bots doesn’t exist yet.
4. So if you don’t have programming skills, the best thing to do is to go to a software development company or somebody who has the experience, right?
Matt Schlicht: Right now, that is going to be the most efficient solution. I think there will come a company and a product that every business can use to get a bot up and running very quickly, and I know a lot of people trying to do that, but that solution doesn’t exist in a truly usable fashion right now. If there is a mechanic trying to put a bot together for his store, or another small business, they’re going to run into a lot of problems with the existing solutions today. Hiring somebody right now would be the fastest and the most efficient way to get a bot that actually does what you want it to do.
5. There was all this hype a few years ago around mobile apps. All software development companies started building iOS and Android apps. Do you think this is also going to happen with chatbots? Is this the new big thing that software development companies will try out?
Matt Schlicht: Bots and mobile apps are very different, for sure, but they both represent distribution channels for a product and for a way to interact and acquire customers. The reason mobile apps did so well is because everybody had mobile phones, and some apps such as Instagram, Snapchat, Vine or Pokemon GO rose to the top and did really well. With bots, most people don’t know this, but more people now are using messaging platforms rather than social networks, which is super crazy. That’s billions of people!
Bots represent the apps of the messaging platform world, so it does make sense that every company is going to need to have a bot strategy. In the beginning, those bots will probably be very simple and straightforward, and they’re not going to be like little virtual assistants and super AI-powered, but they’re going to do something. Over time, in 3, 5 or 10 years from now, with AI, machine learning and NLP progress, the bots will get even smarter and they’ll eventually live on every messaging platform. Therefore, if you want to interact with a business and you don’t want to visit their website, go to their store, or download their mobile app, you could quickly have a conversation with their bot.
6. How can chatbots help companies generate revenue, besides saving them time and human resources by handling customer service?
Matt Schlicht: Those are definitely the biggest ways that they could save money, but there are other ways. Let’s take Facebook for example. I can message my friends on Facebook, but I can also message businesses on Facebook, who are actually receiving messages every day. Unless they have somebody who’s on staff and is responding to these messages, all these messages are going into a black hole. In a lot of cases, that represents missed opportunities for sales, whether it’s in the short term or in the long term.
“Bots can help reduce the cost of customer support, and that’s definitely one way that they can help save money for a business.”
Secondly, if a business receives X messages a day from potential customers and doesn’t respond to any of them, the moment it starts responding to all of them, it’s going to see some sort of increase in sales. These bots are getting better and better, and smarter and smarter, so the percentage increase in sales will also go up over time. That’s definitely the other area where bots will help businesses make money.
Companies could also offer increase bot functionality at a cost, so users could make microtransactions or pay a subscription to access additional features. Advertising could also work on major platforms such as Facebook, Kik, Skype or Telegram, which could include a checkbox that, if ticked, would enable users to include ads in their bots. I don’t know how exactly that would look like, but the same way people monetize mobile apps or websites through micropayments, subscriptions and ads, is probably the same way people are going to monetize bots.
7. Could the increased usage of Messenger bots mean a decrease in the usage of mobile apps? Nowadays, there’s an app for everything.
Matt Schlicht: I think that mobile apps and bots are completely different and that most people don’t have a tremendous amount of apps that they use regularly. They probably have a handful that they use every day, and those apps most likely cannot be recreated through a chatbot. Some examples include Instagram, Snapchat, email and calendar apps, internet browsers, or even Facebook. I’m not going to stop using their services to switch over to something in a bot. These are things that I’ll always use apps for. There could be additional services or experiences that I tap into by using bots. Therefore, I don’t think that it’s going to cut down on apps, and even if it did, it would only cut down on it by only a tiny amount.
8. If AI is too difficult to implement and rule-based conversation flows don’t cover all bases, is there a middle ground for businesses that want a chatbot?
Matt Schlicht: My recommendation is that most chatbots right now shouldn’t have machine learning or natural language processing programmed into them. They should just present very straightforward options. On most platforms, you can even display buttons, things that consumers can already understand through interacting with a website or mobile apps, and not try to confuse them with too much conversational UX. Maybe there’s not even a middle ground right now. I think that you just shouldn’t do much with NLP or machine learning, and you should just be very straightforward, with buttons at the moment.
You can’t use NLP or machine learning right now for two reasons. One is that consumers aren’t used to it, they don’t know how to do it, and their only experience has been with Siri, and Siri doesn’t really work and can be very frustrating to use. Consumers are already preconditioned to thinking that this technology isn’t going to function very well. Usually, when you present someone with a conversational bot, the first thing that they’re trying to do is not use the bot correctly, but they spend the first couple of minutes trying to screw the bot up. They yell at the bot, they make fun of the bot, they try to ask the bot unrelated questions, and most bots aren’t prepared to answer these, and the consumer is left with a bad experience.
The other reason this isn’t very good is that the technology just doesn’t work very well. NLP can’t handle any possible scenario. Imagine that Chipotle had a chatbot, and they were using NLP. Chipotle has had a lot of problems with people finding E. Coli in their food, which is really-really bad for business, but the company has denied these accusations. So if I was talking to Chipotle’s chatbot, imagine what would happen if I tricked it into telling me that there was E. Coli in Chipotle’s food. The only reason I could trick it is because NLP isn’t perfect, and I would probably be able to do that. Then I could go out to all the news outlets and I would tell them: “Look, I talked to Chipotle’s bot, they told me that there’s E. Coli in their food.” CNN and everyone else would pick that up, and that would be really horrible for Chipotle. The only reason that could’ve happened is because NLP doesn’t fully work yet and you can still trick it.
9. What does the future of chatbots look like to you?
Matt Schlicht: Right now, everything is in the discovery phase. People understand that chatbots are a big opportunity, there’s a massive amount of people who can use chatbots, but nobody really understands how to create a good chatbot. Nobody really knows how people are going to use chatbots, and what people want out of chatbots. At this point, everyone is just creating a lot of different things, and nothing is really taking off yet. I think one of the interesting things right now about chatbots is that if you’re a really creative person, you have a really good shot at creating a chatbot that’s better than everybody else’s because the differentiator isn’t technology, but creativity and implementation.
“Bots represent the apps of the messaging platform world, so every company is going to need to have a bot strategy.”
The first phase of chatbots is a discovery phase where there’s a huge rush of people coming in, nobody knows what they’re doing, everybody is just kind of shooting in the dark, and it’s driving competition on the creative level.
The next stage involves very simple bots, that are also very creative. Some people are going to figure out how to create bots that are technically complicated, but are very creative. Those will be wildly successful.
[ebook06]Learn about the history of chatbots, the most relevant messaging platforms, innovative use cases and best practices from international companies![/ebook06]
The third phase, over the next couple of years, will include companies focusing completely on AI and NLP that provide APIs. At a certain point, that technology will become super useful and will work correctly. Everybody that has been building bots won’t have to build their own API, AI or NLP, they’re just going to use a third-party provider, and they’re going to incorporate that API into their bot. So the chatbots resulting in the third phase will be the ones from the previous stage, now powered by AI and NLP. What’s most likely going to happen there is that the bots that were wildly successful as simple bots might not be as wildly successful once you combine them with AI and machine learning.
During the last phase, there are going to be these cool breakout bots that are powered by AI and NLP and they become wildly successful, and they will be completely different from anything from the previous stages. I have no idea what that’s going to look like, but I’m super excited about it. On top of that, it’s really hard to tell when that is going to happen, as it could be anywhere in 3 to 10 years from now. The technology will get increasingly better, but it will take a while until we can use it to its full potential.
About Matt Schlicht
Founder and Editor of Chatbots Magazine, YC Alum, Forbes 30 Under 30, product at Ustream for 4 years (sold for $130mil), advisor to SocialCam (sold for $60mil).
About Chatbots Magazine
The #1 Place to Learn About Chatbots, AI, NLP, Facebook Messenger, Slack, Telegram, and more.
Want to learn more about the future of chatbots and how they are changing online interaction for both individuals and businesses? Read the interviews on our blog or download our eBook.
- Barbara Ondrisek, Like a Hipster: “When You Stop Developing Software, It Instantly Dies”
- Thomas Schranz, Blossom: “Chatbots Transform the Customer Support Experience in Every Industry”