Airport travel-tech company · Travel & aviation
An AI agent that turns an airport app into a conversation.
An AI assistant that understands what a traveller needs and helps them do it (find food, apply the best offer, redeem reward points, get around the airport) without hunting through multiple screens.
- AI agents
- Tool calling
- Conversational UX
- Commerce
Built while working at an airport-travel technology company. The objective was an intelligent assistant that could understand a user’s requirements, guide them through airport services, and help them complete tasks that would otherwise require navigating multiple screens, applications, or service providers, making the airport experience more conversational and personalized.
The challenge
Too many tasks, too many screens.
Airport users often need to perform several different tasks during a single journey:
- Find restaurants and food options
- Order food before reaching a location
- Discover available offers and coupons
- Apply the most suitable available coupon
- Redeem loyalty/reward points
- Navigate airport facilities
- Understand available airport services
- Get answers about their journey and airport facilities
- Receive personalized assistance based on their requirements
These capabilities traditionally exist across different parts of an application, creating friction for users.
The goal was therefore to introduce an AI agent that could act as an intelligent interface between the user and the underlying airport services.
Solution
Ask for what you need.
An AI assistant capable of understanding natural-language requests and converting them into appropriate actions within the application. Instead of navigating through multiple screens, users could interact with the assistant conversationally.
For example: “I’m hungry. Find me something to eat and use my best available offer.” The agent could interpret the request, identify relevant food options, check applicable offers, and assist with the ordering workflow. Similarly, a user could ask about their available reward points or how to use them, and the assistant could guide them through the relevant process.
Key capabilities
What the agent can do.
01
Conversational airport assistance
A natural-language interface for airport-related questions and services. Users can ask questions conversationally rather than having to know where a particular feature exists within the application.
02
Food discovery and ordering
The agent assists users in discovering food options available through the airport application: user request → understand intent → identify available restaurants/items → check availability → identify applicable offers → assist with ordering.
03
Coupon and offer optimization
Rather than requiring users to manually search through multiple coupons, the system could identify applicable offers and assist in applying the relevant discount during the ordering flow.
04
Reward point assistance
The assistant helps users understand and use loyalty/reward benefits, guiding them toward the appropriate redemption workflow. Users could ask:
- “How many points do I have?”
- “Can I use my reward points for this order?”
- “How do I redeem my points?”
- “What rewards can I use?”
05
Airport guidance
Designed to act as a digital airport guide: users could ask for guidance about airport services and facilities instead of manually searching through the application.
Technical architecture
An orchestration layer, not a black box.
- User
- AI agent
- Intent & context understanding
- Tool / API selection
- Backend services
- Response / action
The AI layer was responsible for understanding the user’s request, while application APIs and backend services handled the actual business operations.
This separation was important because the language model was not expected to independently perform business operations. Instead, it acted as an intelligent orchestration layer over existing services.
Example interaction
From one sentence to a completed order.
User
“I want to order food and use the best coupon available.”
Agent workflow
- 1.Identify the user’s intent as a food-ordering request.
- 2.Retrieve available food options.
- 3.Identify applicable offers/coupons.
- 4.Determine which available promotion can be applied.
- 5.Present the relevant option to the user.
- 6.Continue the ordering workflow.
- 7.Confirm the action/result.
Engineering considerations
More than an LLM wired to a chat box.
- Reliable API/tool integration
- Maintaining conversational context
- Intent detection
- Authentication and authorization
- Handling failed API calls
- Validating actions before execution
- Preventing incorrect transactions
- Managing user-specific rewards and offers
- Providing useful fallback responses
- Maintaining a clear separation between AI reasoning and deterministic business logic
Outcome
From features to a service platform.
The project demonstrated how an AI agent could transform an airport application from a traditional collection of features into a conversational service platform.
Instead of users having to learn how the application works, the assistant could help them discover and use the application’s functionality through natural language.