Talk Less! Chat More!
How do you know if you need Chatbot for your Business? The moment you feel that your customer support has failed to satisfy your customer.
Why is it so that customers often slam the companies for bad customer support. It is not for the bad customer support, rather deficit in human resource to solve user queries on demand. Humans are more intellectual compared machines but machines have their own upside — Robustness
In order to solve user queries on demand, one would need intelligence in addition to robustness. Machines can provide the former as well as the latter in the form of assistants called Chatbot.
Chatbot is a conversational agent that you interact via a chat interface. Simply, they are assistants to solve user queries. And moreover chatbots learn eventually. They learn from the past experiences ‘E’ on specific class of problems (tasks) ‘T’ and solve it with performance efficiency ‘P’.
Here’s a sample.
Chatbots can understand customers through Natural Language Processing(NLP), percept through Artificial Intelligence(AI) and last but not the least learns using Machine Language(ML) and the best part is that it can put a smile in your customer’s face at the end of the day.
Evolution of Technology in customer support
Since the dawn of the 20th century, words such as AI, ML has been buzzing out. Here is how technology driven customer support was and is now,
➔ Till 2011 — Internet Driven
➔ 2011–2016 — Data Driven
➔ 2017 — Future — AI Driven
One can do Magic with AI but what it does to customer support is…
So, once you’ve decided to get a chatbot into action ASAP, you need some dirty hands who are experienced at doing so. Say Hello to EggHead.
EGGHEAD — An intelligent assistant
Egghead is an intelligent assistant which helps you in remembering things. The user can ask it to remember things such as your GirlFriend’s birthday, your SSN number, your next meeting and so on. The user can converse with the bot to query what it has remembered. One can also ask the bot to forget things.
Leveraging Natural Language Processing(NLP), we had built a cross-platform application to chat with the bot. The Best Part? Its tightly integrated with Alexa, meaning you can have verbal conversations and also on the go using the mobile app.
What made us do EggHead?
It was a fine day for one of our teammates until he got slammed for forgetting his girl’s birthday. He thought it would be nice if he had an assistant to quickly remember stuffs. This is how the idea of EggHead was sparked and he also knew that it’s been a problem for many people. So he came up with an idea to build a conversational personal assistant.
How was it built?
The bot has an Android and iOS through which the end user can chat with the bot. Moreover, an Alexa agent was developed to allow verbal conversations.
The Tech Stack includes Node and Stanford NLP is used to process the user text. The challenge was to store the relationship between the context to be remembered with the associated question. In order to overcome it, we used Cayley, an open source graph database and MongoDb was also used as the backend datastore.
The NLP processor primarily does two things -To extract the triplet from the user text then it has to handle then has to do Stemming and lemmatization. The result is then processed by the server.
A small video demo is here for you:
The next interesting application was developed by Rigor is the HAL at “SaveTheHacker”, 48hrs hackathon organized by Freshdesk.
HAL is a self learning AI-assisted customer support. HAL provides enhanced customer support experience with self learning AI assistance to handle L1 queries automatically.
Features of HAL:
· Automation of 50–60% repeated task
· Echo/Voice driven support
· Support through WhatsApp
· Customer care monitored
· Self learning AI assistance
· Generative model
Did we mention WhatsApp? Being used by almost everyone, it is tedious for user to download an independent app for each consumer service we tend to minimize the effort by providing support directly through WhatsApp.
We built an auto responder for WhatsApp to handle user queries. The end user can chat with the customer support using WhatsApp nunber of the Customer Support and at the support end the messages are handled by the server used by the application.
Tech Stack
The FAQ’s by the company as well as recent chat of the employees with the customers are fed to the system to train a model. The Messaging Server was built using Node and the conversational agent was built using Chatterbot in Python.
An Angular based SPA was developed for the Customer Support team through which they could see the statistics as to how efficiently they could work with the help of AI. The Support team can also check the recent chat by the AI with the customer.
A Native Android app was developed for the Customer Service to handle the queries from the user automatically. If the Confidence level of the response generated by the AI falls below 60%, the query is transferred to the Customer Support Team to handle it.
The Performance benchmarks
Web application for the customer support:
And atlast…
About Rigor
We are a Design Driven and Innovation focused Technology Company building World Class products for leading brands and startups.
Visit us at http://rigorsolutions.com/