Chatbots are being touted as the new killer app. They are programs that can hold a conversation with a person on any of the text messaging platforms – Facebook Messenger , Telegraph, Slack etc. Since messaging has recently overtaken the use of social networks in terms of usage, it seems that businesses will use the opportunity to deploy automated chatbots into these environments. To this end a number of new chatbot platforms have appeared, the highest profile being Facebook with Facebook Messenger API.
While chatbot technologies are being heavily promoted, the reality of using a chatbot can be frustrating. The first generation of these bots is based on building sets of rules, and if a user asks a question outside this rule set the chatbot can only say it doesn’t understand. The next generation of AI powered chatbots is starting to make an appearance, and these have the useful property of learning as they are used. Obviously an AI chatbot cannot be launched into the world until it has gone through a large amount of training.
The application of chatbots is generating excitement in many businesses. Fully developed and trained chatbots can act as text based sales assistants, can man help desks, deal with customer support, and of course they can support the creation of wholly new businesses. This is all predicated on the growing popularity of messaging platforms, and this in turn represents a return to one-on-one personal communication, albeit one of the participants is a chatbot.
A large number of services exist which support the creation of chatbots. These include:
api.ai – voice enabled chatbots.
Chatfuel – AI based chatbot creation.
Converse.ai – chatbot creation for multiple channels.
Gushup – chatbot creation for wide variety of platforms.
Howdy Botkit – an open source toolkit for creating bots.
IBM Watson – broad capability language processing.
motion.ai – support for diverse user interfaces using visual tools.
textit.in – visual tools for creating messaging dialogues.
wit.ai – recently acquired by Facebook.
Building chatbots is one approach, but many businesses will simply want to buy one and train it to operate in their domain. To this end Botlist offers an opportunity for developers to sell their bots and for businesses to purchase them.
The future development of chatbots will see them exhibiting more intelligence, and we can expect to see tight integration between AI and analytical applications. This will allow a sales chatbot to suggest product recommendations, and a help desk chatbot to diagnose problems. Some chatbot platforms already support voice and this will obviously become a common feature, although simple text is a very popular medium of communication.
There seems to be little doubt that chatbots will become big business, provided the technology lives up to the expectation. AI offers the tools to make this reality, but the old adage that something must be easier, faster and cheaper, if it’s going to be used, means that chatbots should offer a seamless flow of conversation. Anything else, and they will be rudely ignored.