Recent advances in machine learning and artificial intelligence have created an environment ripe for inventors and entrepreneurs seeking to create automated chat engines which can almost mimic human behavior. Since Microsoft first released Clippy, their Office Assistant, professionals everywhere have been realizing the beneficial potential of a virtual assistant. Nowadays, most prominent websites feature some kind of automated customer service or chat service, which helps customers get responses faster and more precisely. This article is aimed at e-commerce and digital executives and will try to cover a rapidly expanding field of AI Chatbots from different perspectives.
What are Chatbots?
Providing customer support service and assistance is an integral part of many businesses, ranging from banking to educational institutions. Traditionally, customer service representatives’ daily activities include monitoring phone lines or live chats for customer inquiries, and providing assistance when customers reach out to them. However, a lot of these requests are routine and involve looking up order statuses, account balances, and other transactional data. Labor costs for maintaining a sufficient pool of customer service specialists are considerable. Large employee turnover and ability to maintain the desired level of customer service is challenging, especially for seasonal businesses. If an automated system could assist customers in at least some of their requests and help call center employees provide better service, wouldn’t it make sense to adopt such a system in an effort to save money and improve performance?
Chatbots are automated chat engines that augment their human customer service counterparts and handle some of the volume of incoming customer requests by focusing on predictable and routine inquiries. Working through both visual and voice user interfaces, they can respond to questions directly through messenger-like systems, as well as through traditional phone line based communications. For the most part of the last decade, since Chatbots began appearing in businesses, they have been using a scripted approach. This involves creating complicated logical workflows, based on expected customer interaction, and pre-recording all possible answers which the system can respond with to each customer request. In other words, they are very sophisticated and smart answering machines, but with the introduction of AI, that is all changing now.
Introduction of AI
The biggest challenge with creating a relatable Chatbot has always been in adding the human experience to any interaction with the machine. Even the most complicated scripted Chatbots can only use logic which they were programmed to do. If a customer asks it a question that it doesn’t understand or is not part of the script, the only solution is to connect them with a real human person. This happens quite a lot, often undermining the benefits of an automated system and making the experience frustrating to the customers. With recent advances in Artificial Intelligence, it has now become possible to humanize the robot. In both, spoken and written communications, Natural Language Processing (NLP) technology can now understand complicated lexical and grammatical structures. Unlike traditional scripted or even Machine Learning based approach, which looks for keywords in a conversation, a Chatbot equipped with NLP will form a comprehensive logical context for what the conversation is about and will then infer the meaning based on human-like understanding. While this approach is still far from perfect, it provides a much more relatable experience.
On a more basic level, advances in Machine Learning technologies have also made these systems a lot more robust and efficient. Even without an understanding of natural human language, modern Chatbots are now able to recognize multiple variations of spoken and written words, as well as their synonyms, all in order to better understand what is being said. Chatbots used for the knowledge base and technical support are also able to match user requests, using an existing database of previous customer issues, with those that are similar in nature, shortening the customer service cycle. Other recent developments in Chatbot technologies have allowed for easier creation and deployment, using pre-existing interaction protocols. Personalities, which also use AI, have become a major component of the Chatbot ecosystem. Customers prefer talking to a machine that has a character, like Microsoft’s teen-spirited Toy.
What Can a Chatbot Do?
This is really a rhetorical question, as almost any function which involves interaction with a customer can be done by a Chatbot. Over the years, Chatbot technology has evolved into ever more complex systems, where computer responses have been modeled using human linguistic semantics and heuristics. Using AI to predict responses expected from a user, allowed the conversational capabilities of these systems to expand, putting bots as the first line of response to customer service inquiries. Many websites now feature specialized Chatbots which know about current product inventory levels and order statuses. They help shoppers find and purchase products online reducing the amount of time they spend shopping. Support and knowledge base access have also improved. Rather than performing archaic and complicated searches through lengthy help files and product documentation, a smart Chatbot can understand natural language and respond with the right information, saving tons of time and reducing frustration.
Another set of Chatbots is aimed at improving productivity by helping with everyday chores. These virtual personal assistants can find the latest news, commute information, and weather forecasts relevant to each user. There is even a bot that can help you find the best place to each and make a dinner reservation at a restaurant, based on your personal preferences. With a wide range of possible applications, the potential for what Chatbots can do is truly enormous.
Platforms, Contexts, and Assistants
Let’s take a quick look at the Chatbot ecosystem. Traditional integrations typically involve custom designed systems that function on individual sites or standalone applications and are tailored to specific business needs. Most Chatbots currently on the market are very specialized, in other words, they are only able to perform tasks and conversations related to the business or tasks for which they were designed for. These bots live on individual websites and function within ecosystems of the business that created them. More recently, there has been a shift towards a more persistent Chatbot presence, which includes additional communication channels. In addition to being able to communicate with customers through online tools and phone lines, these Chatbots can now also utilize other common channels, like instant messaging platforms which include Skype, Facebook Messenger, Telegram, and others.
Another category of Chatbots includes what is called a Virtual Personal Assistant. Apple’s Siri was one of the first such systems, always available on users’ mobile phones. Virtual assistants provide a much broader range of tools and skills, which help their users perform everyday tasks. These general purpose bots can help look up stock prices, weather forecasts, add items to a shopping list, call someone, and even play music. Most major technology players have joined this foray by creating their own virtual personal assistants. This includes Google’s Home Assistant, Amazon’s Alexa, and Microsoft’s Cortana. Traditional Chatbots can also integrate with them, providing their bespoke functionality as part of a broader range of skills available through virtual assistants.
Future of Customer Interactivity
Over the last few years, Chatbots have started to become more ubiquitous and are now expected by users whenever they interact with a business, whether it’s online or through another channel. In addition to providing support and service functionality, there has also been a shift towards more proactive engagement which includes new business development and sales. Combining this with the widespread use of multiple communication channels, as well as the growing adoption of virtual personal assistants, a new trend emerges. In addition to direct communications through online presence and phone lines, customers are now expecting businesses to respond to their inquiries on social networks, instant messengers, and personal assistants. The concept of a persistent virtual identity presents a new approach to customer service, which allows a business to ensure, that their brand image and reputation is consistently delivered across all points of interaction with its customers. Business executives should consider including this concept when evaluating its Chatbot strategy. The following is a very abridged listing of some of the more popular Chatbots, as well as tools and platforms used to create them.
Personal Productivity Chatbots:
- Skype Bots, which includes Scoop – with the latest news, Skyscanner – which can find cheapest flights, and StubHub for latest concert tickets.
- Messenger Bots, available through Facebook Messenger, offer a large variety of features, like news, sports, weather, and entertainment.
- Some standalone bots include NewsBot – a powerful bot that suggests which news articles you should read based on your past reading history, or Replica – which is available as an app on your phone and can be trained to become a virtual copy of yourself.
General Personal Assistants:
- Google Assistant – general purpose assistant, available through iOS and Android phones, Chromebooks, TV sets, Google Home devices, and Google Allo messaging app. This is an open platform which boasts over 1 million third party tools in addition to built in standard features like scheduling and playing music.
- Microsoft Cortana – a more recent entrant to the virtual assistant market, available on Windows PCs, mobile phones, and integrated speakers. This platform has a smaller selection of skills, ranging in 100s, and boasts of being able to follow you regardless which device you use.
- Amazon Alexa – this, more popular virtual assistant is only available through Amazon Echo devices. The skills marketplace contains hundreds of games, productivity tools, and smart home control integrations.
- There are also a number of smaller, but powerful virtual assistants, each with a set of its own unique features: Briana, Jibo, Cubic.ai, and many others.
Chatbot Creation Tools:
- Mycroft.ai – is a new player on the market who create a first open source Chatbot platform. In addition to having basic virtual assistant functionality, this platform can be augmented with custom developed functionality and integrated into many existing platforms. Over time, it is expected to grow into a full-featured personal assistant which would be able to compete with the likes of Alexa and Cortana.
- Conversation.one – a truly cross-platform Chatbot creation tool. The platform allows for GUI based creation, as well as custom code, and can deploy to all major platforms like Alexa and Facebook Messenger in addition to standalone websites.
- Recast.ai – a trainable, NLP bot, which can be extended to include any customer conversational logic. Works on most messaging platforms, such as Slack, Kik, Facebook Messenger, and others.
- Botpress.io – a framework and an agency for the creation of standalone Chatbots using an open source solution. This means that an in-house development team can also get the job done.
- Chatfuel – a popular Chatbot framework which integrates with most messaging and social media platforms.
- BotMakers – a Chatbot developer marketplace where you can find a partner to help you build your next virtual assistant.
- Rebot.me – a GUI based Chatbot creation tool. Has quite a few templates and personas to choose from before adding this to your site.
- IBM Watson Conversation – a Chatbot building platform utilizing Watson AI.
- Facebook Messenger Bot Framework – a platform for communicating with one of the largest ecosystems of users.
- Microsoft Bot Framework – cross-channel bot building framework utilizing Microsoft development platform.
A more comprehensive list of authoring and development tools.