Platform for Live Agent Connection
The Rake bot was created for W5Golf – the company which provides a system for optimization of relevant engagements and improving services. It also provides customer experience optimization solutions and helps develop customer experience strategies that deliver results.
The main goal was to build a bot for not only communication. The bot understands all kinds of questions related to company services and recognizes human speech. Also, the system provides bots hosting and live agents connection. The system is really efficient such as current clients conversation is able to be saved. It helps the bot to move towards the right step when it gets the client’s message.
Duration: 1.5 year.
Node.js and Angular.js developers, QA Engineer.
Facebook – the system automatically adds new FacebookApplication configurations and gets inbound messages from Facebook page he/she is signed.
SMS Twilio – the service that provides clients and live agent communication via SMS and provides notification mailing for appropriate clients.
We created own web-widget to place on any website. Users get access to the bot via mentioned webwidget and messages history.
We selected a set of tools matching our project requirements and client needs.
Botkit is open source platform were used to create bots for different Messengers like Skype, Facebook Messenger, Twilio and Web-widget.
Loopback is the framework that was selected to automate API generation for the work with a database. It includes standard CRUD methods generation and provides delimitations of database access.
Angular.js is an open source framework that allows creating web-interface for administration dashboard for chatbot management.
MongoDB was used for saving clients and bots data.
Google Engine is the platform for project deploying. It provides automation scaling depends on server load.
The first step was prototyping the system that provides bots connection and disconnection without overloading.
Then we created module files that contain the logic of bot’s workflow. After that, we developed an administration panel for customers.
As the system is multilevel, we founded the solution that provides bots addition and removal without system restarting. One of the client’s requirement was system scaling. So, it seems to be impossible to set up a standard Google App Engine balancer. It can break the wholeness of a client’s conversation.
We built the solution that saves the current a client’s conversation and moves the bot to the necessary next step when it gets the message.
- Technical result: We provided automation scaling system that acts depend on server load.
- Business result: We didn’t build a simple bot where are few options for choice on buttons. Our chatbot is able to communicate with people, understand all kinds of questions related to company services and recognize human speech.