Building a virtual assistant is now as easy as playing a flash game and as fast as cooking instant noodles! Today, we show how to build a simple FAQ bot that requires no code!
Several large enterprises use a set of frequently asked questions to support agents. However, adding these manually one by one along with training data is tedious. We’ve solved the problem by introducing a Q&A Hub using which customers can upload a CSV file with a set of questions and answers and train the bot.
Fallbacks are used by the bot for self-learning. You can learn more about the fallback here.
Previously, we discussed Deep Learning, Machine Learning, and Natural Language Processing. We also learned that our NLP engine works on something known as the transfer learning in ML which is how it operates on little data, yet presents accurate results. Today, we show you how to build your own virtual assistant in just 2 minutes.
The bot is trained on a transfer learning model as mentioned before. Transfer learning is to apply the derivations from one NLP task to another.
Building a bot is easy and quick. Even if modifications are required.
It requires little data to work on. How this works is that primary modeling of sentences into similar categories allows the bot to handle small variations easily as seen in the video as well. E.g. Who made you? Who built you? These are two similar questions with the same answer.
The bot is trained on something known as similarity search. The fallback allows the bot to learn about what users are asking. Even the slightest variations in a question are easily picked up
These questions can be easily added back to the bot.
This is especially beneficial for SMB’s and startups that don’t have enough data to build an SOP. The bot can start functioning almost immediately and be modified just as easily.
Would you like to see the bot function for a specific use case? Get a demo.
This further helps us with the following –
Consolidation of multiple data sources – We can now connect multiple QnA data sources along with the Knowledge Graph to improve our resolution rates.
Intelligent recommendations and self-learning – This addresses the problem of infinite input and ensures that the user is always directed to the most relevant answer even if their query is ambiguous. The ‘did you mean’ responses selected by the users are recorded and further used to improve the bot.
Faster time to market – With Q&A hub, you can really go live within minutes, as we showed in the demo. Enough said!
Business user-friendly – The entire process, including training and recommendations, is automated. All you need to do is hit ‘Train’ for the bot to work. Absolutely zero code!
Aavi Mathuan is a marketer, web dev and content creator at Yellow Messenger. Her core interests lie in the space of technology and brand psychology. Her natural affinity towards design, content and storytelling is refreshingly perceptive and insightful. She lives with her happy puppy in Bangalore, India.