Chatbots & AI
Chatbots have been around for a number of years, however a number of global trends are propelling artificial intelligence to go mainstream. According to senior software executives we’re now in the midst of a once-in-a-decade paradigm shift. Messaging is the new platform and bots are the new apps. The shift will radically change the end-user experiences, business models, how we monetize and how we advertise. Chatbots are increasingly creeping into our personal lives and this is being fuelled by giants like Microsoft, Google and Facebook.
A chatbot (or short for chat robot) is a computer application that maintains a conversation with a user in natural language, it understands your intent, and sends a response based on rules and the data and the organisation. A number of messaging platforms now allow for chatbots to be used. This includes Facebook chat, Skype, Slack for enterprise, Whatsapp and even SMS. Artificial intelligence is usually embedded into the application in order to allow it to understand what the conversation is about.
So what have been the trends that have really assisted in taking chatbots mainstream?
Mobile messengers have become mainstream and mobile messenger platforms are now supporting chatbots across the board. Snapchat is the only mainstream messenger platform not currently supporting chatbot development.
Chatbot development costs have been dramatically reduced. A number of chatbot tools now allow you to build chatbots without any prior coding experience. In our follow up posts, I will be assessing these particular platforms and some of the limitations faced by developers. Chatbot builders have been created by a a number of bigger players including IBM, Microsoft, Facebook, Google, and startups and even Open source projects.
App boredom – People are tired of having to download apps and keeping a number of apps on their phones. According to Comscore, as early as two years ago almost two-thirds of US smartphone users did not download a single app within a month and to make matters worse according to Localytics 23% of users abandon an app just one month after using it.
In this particular article, I will be evaluating a number of chatbots. The majority of these; I have used previously and some of them I use on a daily basis. In the comments section I’d like to you direct me to a chatbot that you’re currently using and what you think of it. I’d like to know your likes and dislikes and how they could be improved. If you’ve recently had an idea about a chatbot project you’re looking to build; we’ll be happy to talk about the work.
Ever since, I’ve entered the world of entrepreneurship, I’ve had to be super stingy with my money – my new nickname is Mr. Scrooge at home. I’ve wanted to know exactly where I’m spending my money and how I can take it further. After using my Natwest app and working and categorising my spending manually; it never really got done. Then I came across Cleo whilst having a look at Entrepreneur First – a startup incubator in London. The startups were all trying to solve rather complex problems as the founders came from a technical background. So how does it work; Cleo connects to your bank accounts and becomes your personal financial assistant. It gives you regular reports on your spending; categorising your expenditure into: transport, eating out, cash, groceries, entertainment. If you feel one particular category you’ve spent out of the ordinary on you’re able to drill down further letting you know exactly where you’ve actually spent your money. Since becoming a regular user of Cleo, a number of features have been added which are extremely useful in picking out some of my financial blind spots which I need working on. Apparently, I was regularly spending 93% more than most Cleo users on eating out and Starbucks. As I regularly work from coffee shops, I knew that wasn’t an expenditure that I could sacrifice on. However, my fast food habit was a quick win. Not only would I be saving approximately £100 subtituting takeaways for packed lunches I’d be eating healthier and less processed foods. You might be thinking; why would I want to give away my data to a third party and what if my banking gets hacked? Cleo has this covered; they’re using bank-level security and having partnered with Salt Edge to aggregate the financial data without having to handle sensitive credentials themselves. Salt Edge works with 40 of the world’s largest banks and are compliant with PCI DSS requirements.
The second chatbot I regularly use is poncho. Poncho started out as your personal weather reporter. Poncho comes in the form of a cat, full of personality, everytime it updates you on the weather, it TRIES to make you laugh at the same time. Howeverm with their user growth I’m sure they’re doing a stellar job of making people laugh! Secondly, the weather update is combined with giffy complementing the update. Poncho is also another example of a start up; which started out as a pure chatbot play; but more recently have launched a mobile app which does even more fun stuff such as: horoscopes, mantras, lifehacks and some adorable iMessage stickers. Recently the company founders starred on an Apple TV show called ‘Planet of the Apps’. This was a reality TV competition where app developers and startups initially pitch to four mentors. The mentors included Jessica Alba, Gwyneth Paltrow, Gary Vaynerchuk and Will.I.Am. If the creators manage to successfully win over one of the mentors; the founders go through an intense incubation period before they are ready to pitch to a tier-1 venture capital firm called Lightspeed Ventures. This VC was an early investor in a number of Silicon Valley success stories such as Snapchat and Flixster. It has the ability to invest anything from seed to late stage capital and is known for writing multi-million dollar check sizes. Not only, did Poncho founders manage to win over Gwyneth Paltrow they succeeded in winning over two of the venture partners – Huang and D’arcy. During the incubation period rather than focusing on the artificial intelligence peice; there pitch was geared around Poncho becoming a content platform serving up intelligent alerts, schedules, news and even providing suggestions for background screens in order to monetize the active users.
Niki is an all-in-one ordering chatbot originating from India. It’s mission is to bring humans and machines closer to cash in on the next wave of hyperlocal conversational commerce. It came about – like how every other billion dollar company is formed – a group of friends, over a round of beer came up with the idea of a digital chat-based AI powered shopping assistant. So what does this actually mean? In the last few years we’ve seen a plethora of unicorn startups within the ‘on-demand’ economy with the likes of Uber and Lyft dominating the scene. However, the space has increasingly become diluted with a number of players offering similar services. Niki, plays a unique part by amalgamating a number of these services into an all-in-one chatbot. At the start of Niki.ai’s journey there minimum viable product consisted of only one use case. Users were able to order a taxi using either Uber or Lyft. This was addressing a problem within the on-demand economy. As their user base grew, they started introducing new features. Since the beta launch in June 2015; the bot now allows you to book a number of services including: laundry, a bus, a hotel room, events, movies and are now currently working on flight tickets. Their funding story is definitely worth admiring too. India lacks the investment eco-system that the U.S and Europe has. Scale-ups which have traditionally prospered in the Indian tech scene; have been service oriented businesses such as outsourcing houses which are generally cashflow positive. Building an Artificial Intelligence company at scale requires a large amount of up-front investment and venture capital to build a world-class business.
One of my all-time favorite restaurants Pizza Express have introduced a chatbot into their marketing mix. This particular chain of restaurants love experimenting with innovative promotions, every time I’ve visited the restaurant they gave me some sort of offer to entice me back in. So for this particular brand; a chatbot made sense to gain more bookings and at the same time gain more feedback. Bookings are taken in a rather intuitive manner. It finds out which restaurant you’d like to visit by offering you the option of searching by location and the ‘find nearest restaurant’ captures the geolocation using your device. The rest of the bot is just like any other booking form; it captures a number of your data points to make the booking. The customer service feature is less developed. It directs you to a form on the website where you’re able to voice your concerns about a particularly bad experience you’ve had.
I have to admit there isn’t a whole foods in my local vicinity, so I’m definitely not a regular shopper, but it’s definitely a business which I look up to. Again they’re an example of a real innovator within the food industry; they’re known for exclusively featuring foods without artificial preservatives, colors, flavors, sweeteners, and hydrogenated fats. The whole foods chatbot hasn’t been used to solve a specific use case for the business like booking or shopping. However, it’s been used to engage potential customers and help them solve their problems. Have you ever visited a supermarket after a long day of work, and with so many different things going your indecisiveness kicks in and you just don’t know what to buy? Their chatbot helps to solve this particular use case by allowing you to ‘Search for recipes’ as well as browse for new ones. When browsing for new recipes; you’re able to break it down by ‘cuisines’ or ‘special diet’, followed by ‘main dish’ or ‘appletizer’ etc. Whilst visiting this chatbot I quickly realized that a number of sub-questions could have been added. Such as Asian food could have broken into nationalities etc. Even though adding more features and extra categories could have been exciting the objective of the app should not be forgotten. The purpose of this chatbot is to engage and give ideas. So from a user experience standpoint, it shouldn’t take more than a handful of questions to start presenting the client with the output.
Duolingo bot, is a creation of the Pittsburgh-based language learning scale-up and it’s first major feature update since the launch of the company 5 years ago. In that time period, the company has gained over 150 million users and has persistently stayed at the top of the educational app charts across every available platform the app is available for download. If you haven’t come across the app before, the proposition is simple. The idea is to spend anything between 5 – 20 minutes of interactive training a day to learn a language of your choice. The training takes the form of small tasks which include learning a combination of vocabulary and grammar. In one session you may be asked to translate a sentence, pair a combination of words between two languages and repeat a sentence or phrase collectively. Currently, I’ve been learning Spanish using the app and a number of features have kept me coming back. Gamification motivate me to keep up my daily goals. Firstly, I’m rewarded for hitting my daily goals and I’m encouraged to keep up with my family and friends. According to the Co-founder and CEO – the last four years have been pretty revolutionary. Not only has the app remained free since its inception it has scaled the technology to be able to serve more than 100 million users. It has also continually fine tuned the chatbot by A/B testing. For example should it teach you plurals before adjectives, or adjectives before plurals. In the last year the company has been asking its users more fundamental questions around what does it take to learn a language. A common request they were receiving was the ability to practice a conversation. Friends who are polyglots often complain about language learning apps that they often exclude the human element involved in learning a language. My old school friend, a lifelong language learner – spends months at a time visiting a particular country when learning a new language. ‘I want to be able to converse with locals – be forced to practise, make mistakes along the way or maybe pair-up with a native so we can teach each other our languages’. However, it turns out that this only works well with a small portion of language learners, often polyglots have the confidence to make mistakes whereas others feel too embarrassed to speak to natives. The chatbot gives the app the ability to converse with users without them having to worry about being judged. Currently, it is text-only but in the future their plan is to introduce a fully-spoken version in the future. Duolingo have had to deal with a number of inherent limitations of natural language processing and parsing errors experienced with chatbots. In comparison to the applications discussed previously; the scope of a language teaching chatbot requires the bot to guide you to what you should be saying. So once you’re interacting with the chatbot, the autocomplete feature paced above the keyboard called ‘HELP ME REPLY’ kicks in much harder than normal. This feature gives the user a set of answers to respond to the bot. When answering without using this feature, the bot is able to recognise your response to the closest matching answer when you do make a mistake. The CEO has great plans for Duolingo and his vision is for the chatbot to be as powerful as a real human tutor – “If you had the ability to spend thousands of dollars on a human tutor, our app should be as good as that”.
Creating a chatbot can be an integral part of your brand. if you’re a startup; it can be your sole client-facing application, or it can be an extension of your current app offering. What differentiates your chatbot is the problem you’re looking to solve. For Cleo, their mission is to enforce better financial habits by helping you to look after your money. The value of the proprietary tech will be dependent on how well the application classifies the data to provide its users better insights into their spending. On the other hand; Poncho’s success can be attributed to the quality and vitality of its content. They want communicate the weather, horoscopes and other forms of content in a more engaging fashion. For existing brands chatbots can be utilised for a number of reasons too. Examples above show case how restaurants like Pizza Express are using chatbots to solve a business challenge – bookings. They’ve found a novel way to help people book online so less resources are required to answer emails or their social media messaging.