Rule based or AI bot: which one is the dark horse for your business?

About Webinar

We all want that our bot should be as responsive as possible to our queries. It won’t be wrong to say that natural language is a fundamental requirement of bot technologies. This in turn establishes a direct relationship between the evolution of bot platforms and natural language processing platforms. With the arrival of cloud platforms, the growth and advancement of natural language processing technologies are being carried forward by giants like Microsoft, Google and IBM.

As a result, most bot developers spend time integrating their frontend bot applications with natural language processing services provided by a different platform.

The webinar will cover the two main natural language processing techniques that are quite popular with bot development. These languages are the distinguishing factor between a rule- based bot and an AI bot.

Agenda

• Introduction to bots
• Bot frameworks and natural language processing (NLP)
• Discussion on natural language programming techniques
• Differences between rule-based bot and AI bot
• Demo
• Q&A

Questions & Answers

The following are the answers to the questions that were asked during the live webinar.

Question 1: Can rule-based bots be used alongside NLP to improve upon the question processing?

Answer 1: Yes, NLP and rule-based bots can co-exist, and will provide a better UX for the end user.

Question 2: What kind of budget and time frame do you recommend to develop a rule-based bot vs. an AI bot?

Answer 2: A rule-based bot can be deployed in 2-4 weeks. An AI bot will need another few weeks since the model needs to be trained etc.

Question 3: How do I know which platform is best for my project?

Answer 3: It depends upon the use cases, industry, channels and target audience of the chatbot. i.e., if it is more of a consumer bot for retail customers then Wit.ai and Facebook would be the ideal development platform. However, if you have enterprise users in intranet and security is a major concern, we would suggest MS Bot Framework.

Question 4: How smart are the chatbots that are available today, especially in comparison to humans?

Answer 4: Depending on the use case, there are chatbots that are smarter than humans, but in complex scenarios chatbots are definitely not as smart as humans. Bots are not created to replace humans but to compliment them in repetitive tasks, so that the human can better focus on more critical tasks.

Question 5: What is the market for rule/AI based chatbot in the context of an analytics platform? For example, can it answer questions such as “what are my expected sales figures for the next quarter?”.

Answer 5: The market is huge and continuously evolving. There are many data analytics platforms that have been launched in the market with these kind of capabilities, including Cortana Analytics Suite, Power BI etc.

Question 6: Will NLU work for datasets other than a restaurant or an airline data set?

Answer 6: Yes, NLU will work with other datasets. It supports various other industries and processes such as eCommerce, retail orders, insurance products, etc.

Question 7: What is the future scope of chatbots?

Answer 7: In the near future, chatbots will be able to understand the emotions and the behaviors of users.

Question 8: How do you implement a simple supply chain chatbot?

Answer 8: We can create business logic/workflow in the chatbot and integrate it with your existing backend system with the help of REST Services for fetching data including shipment status, etc.

Question 9: Can a chatbot be used as tech support for troubleshooting issues such as a speaker or a microphone not working?

Answer 9: Troubleshooting simple tasks - Yes a chatbot can definitely help you solve these issues based on its knowledge base. In fact, the chatbot can also be used for various IT support related troubleshooting tasks such as Outlook configuration, network issues, desktop performance etc.

Question 10: Using conversational bots in a health setting can go against HIPPA rules. I don't want anyone to hear about my personal health info. How do you propose to handle these kinds of regulations while building a healthcare bot?

Answer 10: HIPPA compliance can be achieved by making sure that you implement secure and encrypted messaging platform and also if the transmission channel can be encrypted. We can also implement a system where the patient’s identity can be kept confidential.