When Microsoft(Tay experiment), Skype, Google, Facebook (Facebook M) and Amazon (Echo) start to push the needle on intelligent (chat)bots, something is about to change. Not considering it to be so much of a revolution, as rather a strategic direction towards a future in which intelligent bots will continue to permeate our software, leading a new and next generation of personal assistants. A big deal? Hells yea.
While tech behemoths are busy perfecting the technology and distribution models, startups equally experiment with the monetization and many applications of the technology. Bots are at the front and center of this new era with offerings spanning online concierge services (e.g. Alfred, Magic, GoButler, Assist), developer collaboration tools (e.g. Blockspring, Slack, Atlassian), intelligent calendar assistants (e.g. x.ai) and much more.
Botstores vs Appstores
In disregard of purpose of interface, a more generic term would be ‘conversational agents’. Another term recently emerged, called ‘ChatOps’ which already hints at the implications in terms of app architecture and distribution.
As conversational agents will continue to proliferate, they bring additional functionalities and features to a centralized system. Let’s take Facebook Messenger as an example; an application to message your friends. Opening up this platform to external conversational agents, now allows users to book flights, get information on deals, and much more.
The paradigm shifts when you start considering messaging applications as an operating systems, much like a mobile platform, but now with the conversational agents being the proverbial apps.
The implications this has for companies that do not own a successful mobile operating system, like Google or Apple, is that this era of conversational and intelligent agents, hold the key of shifting the value away from mobile apps and push them into instant messaging platforms. And they run cross-platform.
Before conversational agents, you had to open your calendar app on your phone to book a meeting, now you can simply ask chatbot Amy to set up a meeting for you, or ask Kip on a Slack channel to do your shopping. This has been the reason why companies, such as Telegram, Kik, Microsoft and Facebook, have been busy setting up botstores, as well as there already independent botstores, such as botlist or botpages. Again, much like appstores.
See where this is going ?
Bots aren’t always bots
There is a difference between intelligent conversational agents and simple automation scripts that are triggered with a simple keyword.
When looking at the current state of botstores, most bots behave as plugins that users can trigger by using a simple keyword. Examples range from augmenting collaboration platforms, such as Slack, with integrated Twitter feeds, asking weather information, ordering an Uber from your chat channel, or booking a meeting with someone.
The range of usefulness is all over the spectrum. Again, much like the appstore equivalent we know.
New opportunity to turn conversations into conversions
But there seems to be a great opportunity for more intelligent conversational agents, or chatbots that are programmed to engage with a human user. If one-dimensional search queries already convey part of your buying intention, imagine what a meaningful conversation divulges about someone.
Through language we convey our personalities, intentions and aspirations. More than we want to.
Being able to elicit additional information from a human user, through a friendly conversation, enables algorithms to tap into new sources of data that can better profile said user, hereby greatly enhance propensity models and as a result drive higher conversions.
Intelligent bots have indeed a front row opportunity in becoming your personal concierge for a wide range of tasks, and as such have a tremendous opportunity of obtaining instantaneous information you would not disclose in a one-dimensional search query.
Decision making with bots vs. search
Search is one way of retrieving information. We use it to get informed before making a potential buying decision. You want to go on a holiday, so you type in “sunny holiday + discount”. You get thousands of sites where you can book cheap holidays. After you select one, or a few, you make deeper queries on each of their respective date pickers and you get a list of options to compare. Apart from getting distracted by reading articles about Ibiza, Bali or other dreamy destinations, I spent over an hour on this task.
Conversational agents are different breed altogether. When we ask an agent to perform a task, a user already made the decision to fulfill a certain need, but he or she might not yet have chosen howto fulfill that need. Taking my above mentioned search experience, these agents or bots have the opportunity of reducing the number of steps in my decision making process. By simply asking me a limited set of additional questions – when my initial question was not specific enough – and proposing some simple recommendations that I can simply accept or decline the booking process is fulfilled and my tickets are in my mail.
Me: I’d like to book a sunny holiday in June for six days
Stuart: Great, a holiday in the sun. How long do you want to be on a plane?
Me: No longer than six hours
Stuart: Is your spouse or kids travelling with you?
Me: Only my spouse, I have no kids
Stuart: I got a great deal for Dubai between 18th and 25th of June, for a total of 975 Euro at the five star hotel Movenpick. How does that sound?
Me: Sounds great.
Stuart: Do you want me to book it?
This was a rudimentary POC interface (Stuart is a startup at the Startit@KBC incubator) through a simple web form, but we got this down to five minutes between idea and having booked a holiday. Of course, I wasn’t picky, nor was the system calibrated, but it didn’t distract me.
Technically, we could consider bots nothing more to be simple interfaces for launching NLP aware queries to recommender systems. And in a way it comes down to this, but they can be so much more by going beyond the buying decision, and actually automate the fulfillment process of an order.
This obviously requires deep integrations into e-commerce and ordering systems in general. Looking at the early business models of concierge services, they largely first based on a small monthly fixed service fee, but foremost affiliate sales commissions.
Bots can indeed be a great catalyst and an important channel for pushing sales.
Bots are all the hype
With over one billion users on messaging and collaboration platforms on a daily basis, mobile users looking for easy decision-making – sometimes impulse-driven – in an on-demand economy I believe there is a huge potential for conversational agents, or intelligent bots.
Although bots, and chatbots in particular, are not a new item in the tech industry. They have been around for many years, but it seems the necessary platforms are only now finally in place for them to launch into mainstream.
… And they are about to impact e-commerce, customer care, and many other industries for years to come.