Alaska Airlines Natural Language Flight Search Implementation
Company Background and Context
Alaska Airlines, the fifth-largest US airline serving 45 million customers annually, has demonstrated a strong commitment to digital innovation and AI implementation. The company has previously established a real-time data foundation connecting guest data with operations data, laying the groundwork for more sophisticated AI applications.
AI Strategy and Implementation Approach
- Developed an 8-step process from idea selection to production deployment
- Adopted a multi-model approach rather than relying on a single AI solution
- Started with internal-facing opportunities before moving to customer-facing applications
- Maintained strong focus on maintaining customer trust and care in AI implementations
Technical Architecture and Integration
Core Infrastructure
- Modernized tech stack from monolithic to microservices-based cloud-native solutions
- Built on existing Google Cloud infrastructure including:
Gemini Implementation
- Primary use case: Natural Language Destination Search
- Key technical components:
Data Integration Points
- Customer profile data
- Flight availability and pricing
- Destination information
- Historical booking patterns
- Real-time inventory
Development Process and Methodology
- Rapid prototyping and iteration based on customer feedback
- Initially started with chatbot approach but pivoted based on user preference for immersive experience
- Development from concept to production in matter of weeks
- Strong emphasis on testing and validation before customer-facing deployment
Production Considerations
Quality and Safety Measures
- Implementation of factuality checks through function calling
- Verification of flight availability before making recommendations
- Price validation in real-time
- Multi-language support validation
User Experience Design
- Focus on natural conversation rather than traditional search parameters
- Integration of visual elements (maps, images) for improved user experience
- Dynamic content generation tailored to user queries
- Support for complex multi-constraint queries
Technical Challenges and Solutions
- Ensuring real-time availability of flight data
- Managing multiple constraint satisfaction in search queries
- Integrating multiple data sources seamlessly
- Maintaining system performance at scale
- Handling multiple languages and cultural contexts
Production Infrastructure
Core Components
- Microservices architecture
- Cloud-native deployment
- API integration layer
- Customer data platform
- Machine learning recommendation system
Integration Points
- Google Cloud Platform services
- Existing airline booking systems
- Customer database
- Pricing systems
- Maps and location services
Monitoring and Maintenance
- Real-time verification of recommendations
- Price accuracy monitoring
- Language quality assurance
- System performance metrics
- User interaction tracking
Results and Impact
- Successfully launched first customer-facing AI travel search capability
- Improved customer experience through natural language interaction
- Enhanced ability to handle complex travel queries
- Support for multiple languages and cultural contexts
- Integration with existing airline systems and data
Future Development Plans
- Expansion of use cases beyond initial search functionality
- Integration with additional customer service touchpoints
- Enhanced personalization capabilities
- Broader language support
- Integration with merged airline systems (pending Hawaiian Airlines merger)
Key Learnings
- Importance of starting with internal applications before customer-facing ones
- Value of rapid prototyping and customer feedback
- Need for strong data infrastructure before AI implementation
- Importance of maintaining trust while innovating
- Benefits of multi-model approach to AI implementation
Production Best Practices
- Maintain strong focus on factuality and accuracy
- Ensure real-time validation of recommendations
- Integrate closely with existing systems
- Build on solid data infrastructure
- Prioritize user experience over technical sophistication
- Implement robust testing and validation processes
- Maintain flexibility for future expansion
