In my previous blog titled Conversational UIs Coming of Age, I traced back how interfaces have brought Chatbots into their fold. In this piece, I have sketched out the world of Chatbots- from how they are developed to how they would thrive.
What does it take to create a Chatbot? Based on my experience in building bots for Avaamo and Facebook Messenger, I believe there are six key capabilities needed to build a sophisticated Chatbot:
- Connectivity with the Messaging Platforms like FB messenger, Skype, Avaamo etc.
- Natural Language Processing (NLP) to understand what user types or asks.
- Dialogue & context management to continue the conversation intelligently and assist the user to finish the interaction with the Chatbot.
- Connectivity with external systems so that the Chatbot can provide necessary information or takes an action on behalf of the user.
- Hybrid UI response Forms/Cards & menus which makes it easy for users to continue the dialog.
- Business logic and intelligence which is core of the Chatbot.
While platform vendors provide Software Development Kits (SDK) of their own to help with building chatbots for their platform, these SDKs are limited in capability. For example, Chatbots developed using these SDKs can't connect with other messaging platforms. These SDKs also don't offer other capabilities needed. Given that there are so many messaging platforms with Chatbot capability (such as WeChat, Facebook messenger, Kik, Slack, Telegram, Skype etc.,), developing a cross platform Chatbot is not a cakewalk.
Chatbot Frameworks and Services
As they say, every problem is an opportunity for start-ups. Complexity in developing Chatbots has led to start-up boom in this particular area. Currently, there are three types of Chatbot start-ups that are up and running:
- Chatbot framework vendors
- Chatbot vendors
- Chatbot development System Integrators (SI)
Chatbot framework vendors are technology driven start-ups offering capabilities such as Natural Language Processing (point #1), Dialogue management (Point #3) and connectivity with external systems (point #4). Some of these vendors may also provide options to develop hybrid UIs (Point #5). The number of Chatbots framework vendors are increasing by the day and what's interesting is that they've got varied backgrounds and have different approaches. Few key vendors are listed below:
- PandoraBots - Offers Chatbot development and hosting platform. They bring in strong NLP and dialogue capabilities to Chatbots. Developers are required to take the responsibility of connecting these Chatbots to respective messaging platforms.
- Wit.ai - Acquired by Facebook, this Chatbot framework vendor provides end to end capability for bot development. Wit.ai provides several constructs for developers to understand the user intent of the conversation, converse with the user, integrate with external APIs and also integrate with Fb messenger.
- Microsoft Bot framework - Although it isn't a startup, this remarkable piece of art by Microsoft helps develop bots across platforms and languages. Its capabilities include NLP, Dialog management and connectivity to a variety of messaging platforms.
- Api.ai - They are a Chatbot PaaS that provide the easiest way to build sophisticated bots with intelligence. They have been acquired by Google recently. They offer direct integration into messenger platforms.
- Lex - Freshly announced Amazon Lex is a Chatbot service, that is based on the technology behind Alexa. Lex works along with Amazon Lambda and rest of the AWS platform services in a seamless manner
- ManyChat - A Start-up focused on No-code Chatbot development for Fb messenger. This keyword driven bot is simple to develop and could be developed in a couple of minutes
- Gupshup - Very popular SMS messaging platform organization launched a comprehensive multi-platform bot development and hosting PaaS.
- ChatterBot - This Open source project uses NLP and ML to intelligently respond to user questions. The drawback is that it can only help in developing primitive Chatbots such as Q&A bots, which can work on pre-defined knowledge base for use cases like customer service FAQs.
There are many Chatbot vendors, who are focused on selling bots for specific platforms to Enterprise clients. One of the popular vendor is Howdy . Howdy offers Chatbots that can integrate with Slack, to add intelligence to software development, project management and team collaboration workflows. Avaamo, an enterprise focused messaging platform vendor has developed bots that connect to enterprise applications such as Salesforce, Quickbooks etc.
In addition, quite a few mobile application development companies sensed the opportunity to develop custom bots and have geared up for offering Chatbot development services. Some of the notable examples are Botgig and Sumatosoft. These vendors work with Enterprises to develop custom bots.
How does the future of Chatbots look?
Based on the utility of chatbots and the interest that I see with our clients, Chatbots will be the norm soon. Every business will be expected to offer Chatbots on a multitude of messaging platforms. There is a huge opportunity for all the three types of vendors in this eco system, namely Chatbot framework vendors, Chatbot vendors and Chatbot development SIs. I also expect that enterprises will leverage Chatbots as part of their internal and external workflows using platforms like Avaamo. One point of caution though is that, ill designed Chatbots can be annoying, and can risk damaging the reputation of businesses.