ARCHITECTING AGENT VIRTUAL ASSISTANT TO IMPROVE AGENT PRODUCTIVITY AND CUSTOMER EXPERIENCE, FOR A MID-SIZED OR LARGE AIRLINE

BUSINESS CHALLENGE

We analyzed the issues with the current Passenger Service System.

A typical mid-sized or large airline has operations spread across multiple cities, countries and geographies, with hundreds or even thousands of agents being involved in day-to-day activities, such as booking, ticketing, check-in and more. Agents need to be aware of the specific operating procedures, especially security-related, applicable for each country. Apart from the regular tasks, they need to know procedures to be followed during emergencies, threats, attacks or disruptions. Due to these reasons, it is imperative for the airline to have a robust agent reference system that can be looked up by their agents.

Most of the airlines still use legacy mainframe Passenger Service System (PSS) as a common platform that is referred across the globe by all agents. These reference systems need cryptic mainframe entries to view or update data. With the lack of word processing facilities in mainframe, it is tedious to edit, copy and perform other functions typical of any content management system. This limits the airlines in publishing changes to their policies and procedures in a fast and effective manner. With outdated policies and procedures, airline agents find it problematic to handle customers and provide them the incorrect information at times, subsequently leading to customer dissatisfaction and loss of trust on the airline.

Events in the past like the 9/11 attacks and the current COVID-19 pandemic situation have reemphasized the need to adapt to emergency situations, and for information to be disseminated to their agents. This becomes challenging with the current legacy reference system.

THE SOLUTION

We built AWS-based Agent Virtual Assistant

Agent Virtual Assistant (AVA) is a chatbot with simple and intuitive interface providing agents quick access to their airline reference system, reports, and weather details and predictions. Structured based on categories, subjects, pages and page contents, it helps agents to quickly traverse through and reach the page they need.

AVA leverages –

Amazon Simple Storage Service (Amazon S3) to access reference data that is uploaded from a structured JSON file. It provides a simple menu-driven option where the agent can see all categories as buttons. On clicking a category, the list of subjects under the category will be displayed. On clicking a subject, the list of pages under the subject will be displayed and finally the agent can choose a page to display the full contents. Airline can publish the changes to the reference system by uploading the updated JSON file to the S3 bucket. The highly durable, available, and scalable nature of Amazon S3 helps in improving the performance, without any additional management overheads
DynamoDB to handle the varying size of data (JSON file provided by the airlines)
Lambda function to fetch and handle specific elements from the JSON object stored in DynamoDB, as needed
AWS LEX service for the chatbot interface. Based on the utterance or option chosen, it invokes an AWS Lambda function to decide the next course of action. Depending on the options selected, AVA will access the required S3 buckets to get data and dynamically build options to display back to the user
Amazon CloudFront to serve content. It is a powerful tool that helps serve up pages mush faster to users and handle sudden large traffic spikes with ease, all at low cost

THE OUTCOME

Multifold benefits offered by AVA

Improved, engaging experience

Intuitive and easy-to-use interface with simple button clicks

Ability to publish changes to airline policies

Easy replay of previous commands/options to navigate to a different page

Can be easily configured for any content management system which follows a category, subject and page structure

No changes to the code once set-up