Alaska Airlines
natural language search for travel inspiration
Alaska Airlines reports that Alaska Inspires, its genAI destination search, cuts planning time by 75 percent and converts at 7 percent, versus 5 percent for the standard search.
Objective
Open up the discovery phase, long underused on the site, by letting the traveler describe what they want rather than starting from an origin-destination pair, and then connecting them to the inventory and the loyalty program.
The deployment
Alaska Inspires is Alaska Airlines' natural language destination search tool, launched publicly in 2024 and built on Azure OpenAI in Foundry Models. The traveler describes what they are looking for (mood, budget, type of experience), by text or by voice, and the tool suggests, compares, and lets them book. It handles more than 90 languages and connects to the loyalty program to personalize based on points, status, and travel goals. Before this tool, fewer than 1 percent of visitors clicked on the where-to-fly entry point and the average search time ran into dozens of hours.
Results Proof C
Established trade press (CX Dive) relaying the figures from the Microsoft customer story, with a consistent vendor source and a statement from a named executive.
How it works
Documented architectureThe stack in detail
- plateforme Microsoft Azure OpenAI in Foundry Models Generative foundation of the search engine, with support for more than 90 languages and voice input.
- outil Moteur Alaska Inspires (custom) In-house layer that translates the traveler's desire into constraints and queries the network of served destinations.
- infra Connecteur programme de fidélité Personalization of the suggestions based on the member's points, status, and travel goals.
- integrateur Microsoft Platform provider and technical partner for the deployment.
How it runs, concretely
For ops teams-
1Expressing the desire customer
The traveler describes their need in natural language, by text or by voice, in one of the 90 languages.
-
2Interpretation AI
The engine translates the desire into constraints (budget, mood, period) and searches for served destinations.
-
3Loyalty personalization AI / data team
It adjusts based on the points, status, and goals of the loyalty program member.
-
4Comparison and booking customer
The traveler compares the options and moves to booking.
The traveler's description of what they want, crossed with the map of served destinations and the loyalty data. Without a link to the route inventory and the member's status, the tool suggests destinations that are not bookable or not relevant.
How your customers perceive this type of use
Sourced studiesLes consommateurs n'acceptent pas les chatbots par defaut : 64% prefereraient que les entreprises n'utilisent pas d'IA dans leur service client (Gartner, 2024) et pres d'un utilisateur sur cinq du service client par IA n'en retire aucun benefice (Qualtrics, 2025). L'acceptation se construit sur trois conditions mesurees par Salesforce : savoir qu'on parle a une IA, pouvoir escalader vers un humain, comprendre la logique de l'agent.
Acceptance conditions
- Etre informe qu'on parle a une IA et non a un humain (pres de 75% le demandent, Salesforce 2024)
- Un chemin d'escalade clair vers un agent humain (45% plus enclins a utiliser l'agent IA, Salesforce 2024)
- Une logique de l'agent clairement expliquee (44% plus enclins, Salesforce 2024)
Red lines
- Rendre l'humain injoignable : c'est la premiere inquietude des consommateurs sur l'IA dans le service client (Gartner 2024) et 50% craignent que l'IA les coupe du contact humain (Qualtrics 2025)
- Remplacer le service client par l'IA sans alternative : 53% envisageraient de partir chez un concurrent (Gartner 2024)
Sources: Salesforce 2024 · Gartner 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- catalog of destinations and routes
- loyalty data per member
- descriptive content for the destinations for semantic search
Org prerequisites
- innovation or digital product team
- link to the loyalty program
- personalization governance
Possible stack
- Azure OpenAI or equivalent
- semantic / vector search
- connector to the booking engine
The plan, step by step
- Step 1Index the catalog of served destinations with rich descriptive content (mood, budget, season).Deliverable: Semantic index of destinations ready to query
- Step 2Build the engine (LLM plus semantic search) that translates the expressed desire into constraints and returns bookable destinations.Deliverable: Natural language search prototype over the catalog
- Step 3Connect the route inventory and the handoff to the booking engine.Deliverable: Discovery-to-booking journey tested end to end
- Step 4Run a user pilot and compare conversion and satisfaction against the standard search.Deliverable: Pilot report with compared metrics
- Step 5Add loyalty personalization (points, status) and deploy to production.Deliverable: Tool in production with a conversion and reuse dashboard
First step: Index the destinations with rich descriptive content, then expose a natural language search over that catalog before adding loyalty personalization.
The same pattern, proven in other industries
Etsy
agentic commerce (in-chat checkout through an open protocol)
Hyundai Motor India
conversational commerce and lead-gen agent
IKEA
brand assistant published on a third-party GPT store
Sources
- S1 Alaska Airlines works to simplify travel discovery with AI trip planning tool Established press archive pending
- S2 Alaska Airlines: AI Flight Search Tool for Redeeming Loyalty Points Secondary archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.