Building a Smart Chatbot using AWS Services

Introduction

Customers want an easy way to interact with companies, and those companies want a faster way to get to those customers to provide services before they lose them. There are many ways to expose products, services, and information - a company website is at the core of that communication, so it makes sense to use it to interact with customers. Providing a tool that helps customers have a direct access to somebody in the company is something you want to include in your business and marketing strategies, the Chatbot is one of the best ways to help customers obtain direct information and to engage with the business in a more personal but efficient way.

What is a Chatbot?

It is a software module in a website that allows you to provide on-line chat conversation without the need of a real human being interacting with the end user. It also has the ability to perform different tasks/actions based on the user's inputs.

Why use AWS Services?There are many approaches out there that allow customers to build, integrate or use a chatbot for many different needs. At the end of the day, an important thing to consider when adding a tool such as a Chatbot to your system, is to have a cost-effective and scalable solution that allows you to grow and adapt to a changing environment. Using the AWS serverless services such as Amazon Lex, Amazon Connect, and Lambdas along with others, provides a robust, scalable and cost-effective solution for any type of businesses or applications. Because of their capacity to scale, it allows you to go from creating small and simple implementations, to having a big customer support center that is able to manage any user/customer interaction from anywhere in the world.

The Chabot powered by AWS

The core of the following application is based on three main components, Amazon Lex, Amazon Connect, and Lambdas, all can work together or in a separate way to be able to provide a true customer support experience.

Complete Architecture:

The Chatbot User: The user that request anything from the chatbot
AWS Cloud: All the AWS services that work together to interact with the User.
Third Party Applications: The integration of Salesforce, Email and Mailchimp by using Zapier.

It all starts when the user opens the chat on the website, the Chatbot (based on Amazon Lex) walks the user through the interaction process. In case there are any issues or the user requires further assistance, then the system could automatically transfer the customer to a real agent to continue with chat by using Amazon Connect. If the customer requires further private assistance (example: passwords or pins), then the chatbot can transfer the customer to a voice call with the same agent. The system works with multiple tools and technologies such as the following:

The React Chat Module

The chat tool is coded using React.js library, hosted in an S3 bucket and accessed through Cloudfront. The app interacts with Amazon Lex through API (using the SDK) to be able to exchange information.

Backend functionality based on AWS services

Amazon Lex will use programmatic and AI to provide the chat interaction. The system also uses Lambda to be able to execute certain actions based on the user requirements called “intents”, and store all the answers/information obtained from the conversations into a table in DynamoDB. Amazon Connect allows the user to connect with a real person, and to continue the conversation through the chat with a live agent if need be.

Integrated services and end-user actions

Zapier tool allows, in this case, to easily integrate third party applications to create a complete customer support for any business needs. For the current implementation, the user/customer provides the required information through the Chatbot and then gets stored on the DynamoDB database. Zapier then will detect the change and send an email to the end-user, then update the user's information on Salesforce, and finally on MailChimp. This will allow us to follow up with users later on with the sales and marketing strategies.

Based on functionality the system is splitted into 2 phases:

Automatic assistance phase (Chatbot):

The React chatbot module interacts with Amazon Lex which is coded to reply and guide the user to different flows based on the different user's inputs. Amazon Lex Intents (chatbot possible paths/flows) works together with a Lambda function to take actions such as replying back with data, and/or storing the given data on the database.

Real Assistance phase (live agent):

When the chatbot can no longer answer the questions, or when the user asks for an agent or representative, then a session with a live agent gets opened by using Amazon Connect.

Scalability and future integrationsThere are two parts to this implementation to scale as needed: the demand and the appropriate tools. To meet demand, the AWS cloud provides different services and tools that help grow as needed in terms of capacity, building better flows, integrating other services, adding AI, analytics, etc. With AWS services you can easily scale automatically and the only limit you will encounter is the number and availability of your agents. Zapier tool works along with AWS to easily integrate the most popular tools out there. If in the future the CRM is changed or if an analytics/marketing tool is needed, with Zapier you can easily add it to the current process.

ConclusionIntegrating Chatbot into any system could boost your levels of attraction and increase customer service interaction as long as the correct tools and expertise are used effectively. Using these suggestions, now you can start setting up a Chatbot for your business and improve customer interaction and satisfaction. With AWS, you can be up and running in no time, regardless of the size of your needs - we help you with that!

®CBQA Solutions

Back to home