Chatbot Case Study Results
- Multi-channel chatbot accessible through Facebook and WhatsApp resolves 42 requests per hour
- Supports 14 complex dialogs
- Employs custom entities detection algorithm
- Spanish chat interface with option to expand into English, Portuguese, and more.
- Serves an average of 3,000 users a month
- Overall adoption of DevOps processes and mentality
Our client, a leading international airline, was looking for a high-tech solution to their customer service needs to ensure a frictionless and positive experience for their thousands of customers a day. Through a partnership with PSL, a leading nearshore software development outsourcing company, they were able to leverage advanced engineering concepts and technologies to create a multi-channel chatbot, transforming the way users connect with the company, and improving the customer experience by eliminating wait times and reaching a wider audience.
botBuilder / NodeJS / DialogFlow / Table Storage
The client transports users to 78 destinations in 32 countries across North, Central, and South America, and the Caribbean. With a fleet of more than 100 aircrafts and codeshares with more than 15 other companies, the company helps passengers reach their destinations all over the world in comfort with guaranteed efficiency and quality.
The sheer volume of passengers going from one destination to the next combined with layover, baggage, and special considerations information can make providing high-quality customer service a challenge. While traditionally airlines have employed call centers to serve these needs, our client was looking for an alternative to optimize the process and improve the quality of their customers' experience in tandem.
To achieve these objectives, they tasked PSL with building a chatbot able to respond to specific requests on a variety of channels in multiple languages.
Our client wanted to expand the number of platforms on which the company could connect with their clients, reduce frustrations resulting from long wait times, and streamline confusing customer service channels, providing accurate and timely responses.
The team needed to build a chatbot that would integrate fully with various messaging apps to provide the greatest value to end users. But, in order to build the foundations of an efficient and advanced chatbot, PSL needed to guide the client through the implementation of advanced software engineering concepts, such as continuous integration, continuous deployment, and automated testing, ensuring the chatbot could be maintained and scaled in the future.
The client received a highly effective and scalable chatbot that successfully serves an average of 3,000 users throughout Latin America every month.
When working with a client that's shifting development processes in order to accommodate a new technology, it's essential to manage changes on a technical and managerial level. PSL defined the project scope by selecting the most relevant technologies and engaging DevOps experts to drive change management and process improvements for the client.
The PSL team configured the integration and deployment servers and worked on automating the builds and releases to ensure continuous integration and deployment was achieved, all the while guiding the client team in the adoption of DevOps principles and processes.
Because of the project's structure, the team built an ecosystem of microservices in NodeJS that fully interact with one another. This ensures that one piece of the ecosystem can be updated or changed without affecting the other parts and is less likely to experience cascading failures. Additionally, chatbot requirements are completed in conversation and interactions, not functionalities in front-end or back-end configurations. This makes a microservices architecture essential for adding features quickly and efficiently in the future.
Designed to service airline requirements, it was imperative that the chatbot confirm flight origins and destinations accurately. However, the client had custom rules in place that made it impossible to use Natural Language Processing (NLP). So, when the flight origin/destination text functionality presented a problem, PSL constructed a proprietary entities detection algorithm to determine the origin and destinations of flights, despite the NLP restriction established by the client.
Likewise, the volume of information intake that the bot needed to handle required a database with considerable reading and writing speed. To solve the need for the application to be able to "talk" to thousands of users at the same time, they implemented Table Storage. The technology allowed the bot to run multiple instances and share the same data quickly.
The client received a highly effective and scalable chatbot that successfully serves an average of 3,000 users throughout Latin America every month. It's equipped to respond to inquiries involving ticket fares, baggage allowances, flight information and confirmation, and more. The chatbot has provided guidance to almost 12,000 unique users in its first six months and is poised to solve even more customer service requests in the future.
Step into the world of chatbots! Meet with our experienced team.