In the previous article, we briefly mentioned the how text based communication is going to power business in coming future. Chatbots are the primary mechanism through which this trend is manifesting itself. A correctly implemented chatbot can help small businesses implement a sound customer outreach strategy and create a customer support mechanism without investing a lot of money.


However, there is a need to understand what exactly chatbots are, how they work and the different scenarios they can be useful in. Business can then look at appropriate implementations to match their use cases.

Simplistically put, chatbots are automated software programmes which provide a way for users to send their queries or specify the actions they want to take-for example, check the weather, check the stock or book a ticket, using many text messaging platforms, including SMS, Facebook Messenger, Skype etc. These software programmes are powered by search, Artificial Intelligence (AI), Natural Language Processing (NLP) and Machine Learning to varying extent, depending upon the use case and what kind of bot is developed. Primarily, bot use cases can be divided into four distinct categories, depending upon how evolved they are in terms of providing human feel to the conversation.

Options-based bots

For the starters, there are bots which allow interactions mostly based on pre-defined options. User is presented with a sort of menu card, from which they can pick their choose. Remember the IVR response when you call a support center? This is the text (or voice) version of it. These types of bots provide a navigational flow using pre-scripted instructions & responses and radio buttons & selections. The content itself may have text or rich media formats. These types of bots are useful when you want to restrict the user input to a few known options.

Search-based bots

Search-based bots go beyond the mechanical interactions of the options-based bots. They allow for greater subjectivity & greater human-feel as far the user interactions go. Here the users can enter a more natural expressions and sentences, but behind-the-scene, the bot would look up the results based on the presence of keywords in user input. For example, if you ask the question to an online clothing store’s bot; “show me all available menswear”, it’s going to search on the keyword “menswear”. Though it might seem like it’s responding to a natural language query, it won’t respond accurately if the query becomes a bit complex with variations with sizes, color preferences etc. depending upon how the bot is set up. For such more complicated results, the next category comes into picture.

Pattern-based bots

Taking the above example further, the pattern based bots can understand more complex queries. They can decipher a full-blown sentence when it follows certain structure. For example when someone enquires about “What is the flight schedule from New York to Washington?”, pattern-based bots can understand that the schedule needs to be retrieved for flights originating from New York and going to Washington. These type of bots are particularly useful when there is a huge underlying content, which needs to be searched based on limited parameters. Working with pattern-based bots may feel as if they can understand human interactions of any complexity, but that’s actually not the case. If the pattern does not match, they don’t work.


These are basic types of bots which can get you started and may be able to handle routine operations & interactions. They are complemented by human assistance when deviations happen. However, truly powerful & evolved bots provide real “conversational” experience. This is where companies like Google, Apple & Amazon are focusing their efforts. They understand user intentions through highly unstructured interactions and respond through highly trained datasets. As the bots infrastructure & capabilities evolve, we will see the extent to which these capabilities can simplify our lives.