Creating a corporate travel flight booking assistant using AWS Lex.

Background

Speech recognition and natural language understanding are some of the most challenging problems to solve in computer science, requiring sophisticated deep learning algorithms to be trained on massive amounts of data and infrastructure. Fortunately, the global cloud providers are now solving this problem and offering speech recognition as a service.

The Brief

This project allows corporate travellers to have a speech conversation with a conversational bot to determine flight options for corporate clients. For this project, Fathom chose to use Amazon Lex (the service that powers Amazon Alexa) to build a conversational interface between a traveller and a back-office corporate travel agency.

The app allows the corporate travel customer to invoke the app through speech, supporting search for origin and destination airports and dates and times. It provides the best three options (recommended, shortest duration and cheapest). The app also offers the option of emailing options or booking a preferred option.

The Technology

The application uses Lex, Amazon’s Speech to Text language capability and its converse Polly provide Text to Speech. The system uses Amazon’s serverless technologies - Lambda, API Gateway, Cognito. A well know GDS API is used to search for flight options. As with all Fathom projects, infrastructure deployment is automated and CI/CD pipelines are set up for production and staging systems.