I hear my alarm go off at 7:59am. I always like to set it to a random time to wake up. I’m in San Francisco for the MobileBeat 2016 Conference held by VentureBeat. After a couple of minutes, I peer out the window of my corner room at the InterContinental Hotel and watch the goings-on of Nob Hill. People are walking and cars are racing up and down the hill. The sky is pure blue. It’s a summer’s day in San Francisco. And today, I’m ready to learn about bots. It’s going to be a good day.
Why am I ready to learn about bots? Because bots are the future. Actually, bots are already here. One of the big buzz keywords at the conference was “conversational commerce” and I think that’s a good term.
Conversational Commerce – What is it?
Conversational Commerce is about having a messaging interface where a user can have a more conversation type dialog with a company’s bot to complete transactions. Chris Messina launched a big post on bots in January and since then the landscape has really developed since then with the launch of so many platforms the past 6 months.
I don’t know what you think of when you hear “chatbots”, but before the conference, I thought of a little, annoying Twitter or Messenger account lobbing marketing messages at its followers all day long. It didn’t sound very appealing to me. Then, the first day of the conference happened. One of the first phrases I heard was “Conversational Commerce”. The term was coined by Chris Messina and he walked us through where conversational commerce emerged and where it’s going. It’s very clear that this area of bots and messenger apps has huge potential.
Just seeing some of the people walking demos through a more conversational style within a messenger app to completing commerce transactions was much more natural and effective. Conversational commerce walks through a similar work flow that would happen on a web site or mobile app. It starts with how a humans initiating conversations in any sales context. Users start through a search process and then filter through with preferences and finally complete the transaction.
A lot of people initially argued whether or not bots would take over apps, but it became clear by the end of conference, that that’s not the case. Bots are complementary. Bots and messaging are here to create a more layered and richer user experience with a company, brand or product.
What is a bot?
I would answer that by saying it is an “intelligent, asynchronous messaging interface”. There’s a lot of power packed into those 4 words.
One, it’s “intelligent“. Companies are focused on building representative, conversational, natural layers into their bots. Some are heavily focused on NLP, Natural Language Processing. Some are focused on NLU, Natural Language Understanding — we’ll talk a bit more about that later. The “intelligent” aspect about bots is bringing natural flow, context from bringing preferences, and data from other instances and other systems. The good companies are really focused on building complex intelligence into these systems, so they truly mimic how people operate and communicate. That’s what we initially tried to do with websites in the early days and then with mobile apps, right?
Two, it’s “asynchronous”. There’s a lot of power by building asynchronous systems. Before commerce and content were mostly synchronous. You had to complete the full checkout process synchronously, before the process was complete. If you wanted to do a search, you typed in a keyword and then immediately got specific results. But, allowing for asynchronous communication allows for a lot more power to be added to the system.
Think about Tesla and the power of bringing their batteries into our electric grid system. As it currently works, power plants have to generate power at the exact time of use. They generate the power and it flows across the wires to be consumed in milliseconds after generation. It’s required to be that way because we have no way to store the power plant’s energy to date. Soon, though, we’ll be able to unpack the energy production with the energy consumption, use storage capability and optimize both processes independently. That’s what’s possible with asynchronous communication.
Three, is “messaging”. There is no real graphical user interface with bots, and in many cases, that’s a good thing. We can model how humans operate, which is through conversation with one another. Just seeing the power of having a conversation with a travel bot and how much better the user experience can be for setting up hotel and travel options was eye-opening for this writer.
Four, it’s an “interface”. There was a new acronym for me, CUI — Conversational User Interface. The messaging interface, like in FB Message, KIK and WeChat will be the new presentation layer to build into our main systems. First, in the 90’s and 2000’s it was web. Then, the last 7 years or so, it’s been about apps. The next frontier is bots and messaging as companies add them as presentation layers to existing web and apps. Too, there will be stand-alone systems entirely presented through the messaging interface.
Are Apps Dead?
There was a lot of discussion about whether bots will drive a nail in the coffin of apps. Are bots going to take over apps? And the general consensus is, absolutely not. The 1990’s brought us websites. And the 2000’s brought us web apps. And then a bit later, over the last seven years, has been the age of apps. And now the paradigm is shifting to bots. A whole new level of service and functionality are going to be available through interacting with brands and companies through messaging and bots.
The goal of a good product manager, developer or consultant will be to go through the use cases where a bot will augment functionality to deliver core services.
While this list is not comprehensive, here are some ways bots are being used:
- Personal Assistant – Several companies are building personal assistant bots that will know personal contact, credit card and preferences and will be able to apply them in a series of transactions on your behalf.
- Search – Several bots can find more complex results than what a traditional Google search experience might offer
- IoT – Many bots are looking to integrate the Internet of Things (IoT) and creating a rich user experience to interact with the IoT
- Banking – Several companies have banking bot app to help users with banking type notifications and banking transactions.
- Travel – Several companies are taking on the travel space and making air and hotel search, filter and buy experiences much richer through conversation commerce
- Alexa – Amazon’s NLP conversational product Echo that allows its users to choose music, movies or make any Amazon orders via NLP bot is another example.
- This list is by no means comprehensive
Generally, we see most customer-facing bots in FB Messenger, though, companies were quick to point out they are messenger-platform agnostic, most customer-facing bots work on SMS, WeChat, Kik or whatever favorite messenger app you use. Then, we saw most business-facing bots are in Slack. Slack actually allows you to build bots and then create personas, add them to channels and even export the bot via JSON to be consumed in another messenger environment.
Do You Want To Build Your Own Bot?
Here are a few bot platforms I found that you can use to build your own bot:
- Olabot – is one of the first, simpler platforms is olabot. Chris Messina (messinabot) is a good example of what’s possible in building a personal or company bot. (Level I)
- ManyChat – is probably the next level of complexity allowing you to build a bot and, also have some reporting and admin back-end. This would be an entry level option to build a bot for a company. They allow news, content and notifications to be customized in the bot/messenger platform. (Level II)
- SoundHound – is probably the next and most complex option in building your own bot. This option allows a consumer to customize their bot for 100 domains on their site, while also providing a customized/private domain bot. It incorporates NLU and was one of the most interesting companies in the space. (Level III)
Chat bots – Consumer & Enterprise
Here’s a an overview of the Bot companies, how they are used and which ones I think are the best for what:
- SnapTravel – SnapTravel was the first travel bot company I saw present. It was good for hotels only but showed the full use case of searching, filtering and completing a buy-transaction via conversational commerce all done in their messenger app.
- Mezi – Mezi was the next level in a travel bot. Mezi does a bit more NLP and allows for hotels and air search and transactions. Mezi brings in context from other sources. This seems like a strong user experience inside a messenger platform for all travel.
- Sound Hound / Hound App – Download the Hound mobile app and see one of the best consumer apps for voice-enabled, true NLU interaction and engagement. This one to me, was the strongest consumer messaging/bot app. It can walk through a full travel use case and does it with totally natural language. You can really provide conversational language with additions, exceptions and also allow the users further refinement. I would say this one was the most interesting consumer messaging/bot platform. (Best in Show)
Enterprise / Business Bot
- SoundHound / Houndify – This one has been mentioned before, but this platform has an enterprise / business option as well. Using the private domain option allows companies to build custom bots. Check out more at houndify.com.
- Kore – Kore seems to market itself as a true enterprise bot play. If I wanted to build a custom enterprise bot, with little or no coding, I would seriously look at this platform. You can use Kora, their default bot and customize, you can build a bot and put it on their platform or you can simple buy an existing bot and customize. For more, check out kore.com.
- Slack – So, Slack is in the beta stage with the bot-building platform. A lot of business bots are built for the Slack messenger platform, but you can also build bots in Slack. Then, you might use an integration layer, like API.AI to help augment functionality or extend with other 3rd party services and functionality. This allows you to build, then export bot into a JSON object to work on other platforms.
Platform Bot Companies (Integration Layers)
- PAT.AI – PAT.AI looks to solve NLU, Natural Language Understanding. They go about solving the problem differently than others in the space. You could add pat.ai to your bot to gain NLU functionality. For more, check out http://pat.ai.
- API.AI – This company/product allows you to provide an integration layer into your bot messenger platform architecture. You could add NLP capability or integrate with other services. For more, check out http://api.ai.
- ANGEL.AI – Angel.ai provides for natural language search specifically for commerce. The technology pulls out the actionable intent to drive the conversation forward. For more, check out http://angel.ai.
- Houndify – Houndify seems super interesting because you can search and customize one of their 100+ public domains. But where some of the interesting power in this platform lies is in setting up a private domain. For more check out http://houndify.com.
- Viv – Viv is an AI platform that enables developers to add an intelligent messaging interface to their existing apps and systems. This one seems interesting. For more see http://viv.ai.
There are plenty more companies and products that were presented and many customer companies that have adopted the technology. Overall, in the consumer space there were several mentions of Alexa and I’m sure you’ll be hearing more about that. Check thisBot Page on VentureBeat for more details.
So, there’s a lot to summarize here. I would say, if I wanted to build a personal bot, I would check out Olabot. If I wanted to build a 1st generation company bot, I would check out ManyChat. If I was ready for serious investment, I would look to build on the Houndify platform. And, if I wanted to build a custom, enterprise messaging bot, I would look at Kore or Slack with API.AI.
It was clear at the conference that, if you wanted to build a true enterprise level bot, you needed to look at serious investment. These development projects can be of the size of major web and app development projects. It will be important to consider if you want to build a bot and messaging platform that may be throw-away in a couple of years or if you want to start building a truly scalable, extensible platform for the latest human/machine interface model.
Next topics I’ll be handle are: interaction design and more about conversational commerce. I’ll be writing more on this topic, but I wanted to give kind of a survey view into what I learned at MobileBeat this year. There now, I saved you $700. You owe me.