← Back to Case Studies Consumer AI

FlightsApp: AI-Powered Conversational Flight Search

Serverless flight price comparison platform with natural language search, intelligent caching, and conversation memory.

The Challenge

Flight search is tedious. Users bounce between multiple airline sites and aggregators, re-entering the same search criteria. Traditional search interfaces force rigid date/destination inputs when users often have flexible plans ("somewhere warm in March for under $500").

The Solution

FlightsApp brings conversational AI to flight search. Users describe what they want in natural language, and the AI translates that into structured searches across the Amadeus API, remembers context across the conversation, and refines results based on follow-up questions.

Key Features

Natural Language Search: "Find me flights to Tokyo in April under $800" instead of filling out forms
Conversation Memory: The AI remembers your preferences and previous searches using DynamoDB-backed context
Multi-Layer Caching: Intelligent caching reduces API calls and speeds up repeated searches
Price Comparison: Aggregates results from multiple airlines via the Amadeus API
Flexible Date Search: Supports fuzzy date ranges and "cheapest month" queries
Serverless Architecture: Runs at ~$0/month when idle, scales automatically with traffic

Technology Stack

React 18 Lambda DynamoDB S3 CloudFront Cognito Amadeus API OpenAI GPT-4

Architecture

React frontend on CloudFront. Lambda functions handle API orchestration and AI conversation management. DynamoDB stores conversation history and cached search results. Cognito manages user authentication. OpenAI GPT-4 powers the natural language understanding layer.