Header Ads

IBM Watson Conversations: Building AI Chat Bots with Ease


We have all heard, if not aware of IBM’s cognitive system, Watson. It enables a link between people and computers enabling them to partner mutually. In order to strengthen this symbiotic relationship between human and a computer, IBM has come up with Watson Conversations. This application helps to deploy messaging platform on any device by using natural language algorithms.

The best part about this app is, one need not be conversant in technology but few simple clicks can build, test and deploy bots across messaging platforms, physical robots or even mobile devices.

IBM Watson Conversation is very handy tool for developers as only 6 steps are needed to get started with Watson Conversations.

Step 1: Navigate to Watson Conversation Homepage where you have access to sample apps and documentation. When you have gone through the documentation enter IBM Bluemix.


Step 2: Register on IBM Bluemix and get options to choose between free and standard plan to use Watson Conversations.


Step 3: Once registered, name your service by clicking “Create”. This gives to access to create and name your own workspace.


Step 4: Once you have entered your tooling interface you can start by defining “Intents”. Intents are like metadata that define your services. For instance, the intent of question “how will be weather tomorrow?” could be “weather forecast”


Step 5: After you have had your intents fixed, define entities that will better help in extracting a better answer to the question, for instance, with respect to previous question “tomorrow” is an entity called @time.


Step 6: Once the entities and intents are defined properly, the Dialog Builder will be ready to answer questions.


Once the configuration of your iBot is finished, you should test the conversation with IBM’s testing tool to verify if the job is finished in its perfect condition. You can try out some statements making sure the statements are not exact as the training data provided.



Diagram shows the overall architecture of a complete solution

There are a couple of quick fixes and tips present all over the internet in blogs, posts, forums etc. One such tip is, along with other statements, adding #return in the intent will add a banner in your test tool stating the system is updating “Watson is training on your recent changes”. Once done another banner will get listed on the app stating your changes have been deployed.


Once the chat bot is trained to your expectations you can embed this to an application and connect with other services. This is taken care with Conversation API, if you are stuck somewhere IBM’s support forums, documentation and sample applications are always ready to provide guidance.

We have a plethora of instances running around for Watson Conversation, hotel reservation assistant, pizza ordering, flight reservation assistant, also people have come up with bots that answers to questions of their white papers.


These bots are very useful to work with and have a conversation with. Try out some examples at Watson’s website and you will feel an urge to have a personalized one for your website.



Watson Conversation service enables developers and business users to add a natural language interface to applications to automate interactions with end users across any channel such as mobile, messaging, and robots

You don’t have to worry energizing all your interests around iBots and creating one for yourself. You have an understanding of the opportunity for your business and Watson Conversation have the right skills to assist you in developing your chat bot. Get started on IBM’s website with videos, demos and basic tutorials.

No comments: