Conversational Interfaces and Return of Command Line
My affair with computing started with an IBM PC running MS DOS. I still remember the seductive appeal of a blinking “C” prompt .It was fascinating, type in the right command and magic happened. Although back that there was very little you could have done with computer that was of any earthly use. But that didn’t deter some of us. We used to spend hours typing and retyping cryptic commands like DIR, ATTRIB, DEL, COPY (later on Ls-axf, ECHO, WALL, SED, AWK, GREP in Linux). There was something about command line that was fascinating. Something, which for some reason was missing in the graphical interfaces which succeeded them. I didn’t knew it back then but today with the benefit of hindsight I can say that part of the appeal of command line was the fact that it was very raw and It was not for everyone. That was my superpower .I knew how to communicate with computer. You need to understand that at that age I couldn’t have said the same about rest of the things, which were seductively appealing to me. In the years that followed technology quickly evolved from MS DOS to Windows to Browser to Smartphones to Tablet to Wearable and now with IOT we are talking about putting computing in practically everything. One would assume that command line would be nothing more than a faint blip in the rear view mirror. Well you can’t be further away from the truth.
The Shift:
As I write this world of computing has gone through a generational shift. We are living in a smartphone era. Messaging has firmly established itself as one of the biggest use case of smartphone. Facebook bought WhatsApp for a massive 18 Billion USD. They are still trying to buy Snapchat. Slack: A group-messaging product for enterprise is worth in billions. WeChat and Lines are massive in China. A lot of VC money is flowing in this space. Last time I checked there were 50 startups trying to get a slice of action.
So we all agree that messaging is the next big thing. However what most of us fail to realize is the fact there is more to this trend than just a land grab .We are in the middle of a paradigm shift in human computer Interface (HCI). We are at the start of Conversational User Interface (CUI). As I mentioned in the beginning, command line interface had a learning curve. DOS Commands were very cryptic and it was anything but natural for human to operate that way. This was a big deterrent in widespread adaptation of software. When we moved to graphical user interface we made significant improvement but still one need to be familiar with the notation, icons and workflow. Every software had a different user interface. There were fun, efficient and sleek interfaces but that didn’t change the fact that one need to learn them. Interaction with computer is structured, discreet and constrained. Its transactional in atomic unit of type, click, tap, pinch, scroll, pan, zoom, swipe etc. That’s not how human communicate, Human communicate in a continuous, fluid and seamless format of communication called conversation.
Lets think a bit more about idea of conversation. Most of us that onlypre requisite for conversation is shared understanding of one common language. But there is more to it. A meaningful conversation also requires a shared context between both parties (meaning of “meet me at Bank” when said in the middle of city on workday is different than when you say it at a beach resort during vacation). Second thing required is the ability to retain state while you move from one topic to another (Try asking “what’s the date today? “ to a booking clerk while buying ticket and try doing the same with a software?). There are many such nuances which comes very naturally to human but for machine it’s not that easy. Schematic diagram of a typical conversation processing software will look like this (source: MIT).
Doing this kind of things with software require some serious technical chops. No wonder that first few entrant in this domain were biggies like IBM, Google, Microsoft, Nuance and Facebook. But as it often happens in technology, real fun starts only when infrastructure part becomes a commodity. We are now witnessing that surge where a lot of services are transforming themselves by applying CUI. Some of them pose serious threat to business model of incumbents.
East is The New West:
PC first generation grew up with a mental model where every activity initiates from “Start “ Menu than we hop on to an application. But there is no start menu on smartphone. What will be the mental model of a mobile first generation? If popularity of messaging is anything to go by than preferred mode of interaction for Mobile first generation would be messaging. You simply tell your machine (PC/Phone/Tablet) what you want and it gets done. Sound very futuristic? Not really. If you want to see it in action look no further than china. Look at how they are using WeChat for everything from well. Chatting to hailing cabs to transferring money to buying groceries. In china one App rules all .Now think like a product manager and answer: If you need to design an app that can do everything what is the interface paradigm you will use? Naturally the one which is most native to human: Conversation hence CUI.
There is one subtle but dangerous aspect of this arrangement, which we need to pay attention to. When you take away the complexity of decision making from user you also take away the choice. When user ask you to book a table for dinner. You have the freedom to use Opentable or zomato or Reserve. This is not good news if you are Opentable, Zomato or Reserve. For they become hot swappable. By same logic what’s true for Open table and Zomato in table booking is true for Google and Bing in search and true for Flipkart and Amazon in Shopping. When you are a commodity service hidden from user there is little mean to differentiate and inspire loyalties.
What’s Next?
I believe that China is a petridish of future. People will look at Chinese market and adopt the model according to the constraints of local market. In India you have companies like Haptik, HelpChat, GoodService magic tiger want to be your concierge of choice for everything from Wakeup call to Shopping and from Web Check In to Restaurant booking. They are doing it by giving a simplified and conversational experience to their users. This experience is powered by an IM interface around a Cascading fallback of BOTS + Structured Inline UI elements + HUMAN Operator. On enterprise side startups like Konotor and HelpShift are trying to provide messaging capability as an SDK to every app developer.
Globally Facebook has launched Facebook M which works at the intersection of AI + HUMAN. They have also acquired a very interesting startup called wit.ai. I assume that at some point they will open the API and make FB Messenger next Big channel for Businesses to interact with their users. Revenue potential of this channel is very real. Skullcandy is already getting fantastic result.
There is one very interesting company which can be huge is WAND LABS. Started by ex googlers this company is building the concept of actionable messaging where any app can become a service on their platform and user can interact with that app via plain text messages. Read this excellent review of WAND by Steven Levy. This is one startup, which has a platform approach, has the right team and can be a game changer. Chinese are not complacent either. Companies like Quixey are working hard to enable deep linking in apps so that content of apps can be accessed without having to install apps. Quixey is Alibaba funded. Alibaba has also invested in their own version of Android and a context aware location service. It’s only a matter of time when they will start leveraging them.
By any standards these are early still times of CUI and we have yet to feel the disruption caused by it. But make no mistake battle lines are drawn and bloodshed is about to start. After a decade of hiatus command line is making a come back and this time everyone is invited. Grab the popcorn this is going to be fun.
