Airlines can now deploy agentic AI across their production workflows, from voice rebooking to digital marketing, according to a new report by Amadeus backed by Microsoft, released in June 2026. The report marks a pivotal shift from experimental AI to operationalized, autonomous agents capable of executing complex tasks within the stringent governance and security frameworks of the aviation industry.

Travel technology giant Amadeus, in collaboration with Microsoft, has laid out a comprehensive blueprint for integrating agentic AI—systems that can plan, reason, and act on behalf of users—directly into the critical operational layers of airlines. The report argues that with Azure’s enterprise-grade governance, airlines can finally move beyond chatbots and into AI that dynamically rebooks passengers during disruptions, personalizes commerce, optimizes aircraft turnarounds, and even executes real-time marketing campaigns.

The Road to Agentic AI in Aviation

For years, airlines have dabbled in AI—predictive maintenance, basic customer service bots, and recommendation engines. But these were narrow, reactive systems. Agentic AI represents a leap: autonomous agents that perceive their environment, make decisions, and take actions without constant human intervention. In an airline context, that could mean an AI that not only detects a flight delay but proactively rebooks passengers onto the best alternative, considering their preferences, loyalty status, and connecting flights—all while adhering to regulatory and policy constraints.

Amadeus’s report stresses that the technology is now mature enough for production, thanks largely to the governance, security, and responsible AI tooling built into Microsoft Azure. Azure’s AI Foundry, Prompt Shields, and Confidential Inferencing allow airlines to deploy agents safely, ensuring data privacy, compliance with GDPR and other regulations, and protection against prompt injection or model drift.

The Governance Imperative

Governance is the linchpin. Airlines operate in a heavily regulated environment where errors can cascade into safety risks, massive financial losses, and reputational damage. An AI agent that mistakenly cancels a booking or misroutes a passenger must have guardrails. The Amadeus-Microsoft report details how Azure’s governance framework provides the necessary controls: role-based access, audit trails, content filtering, and real-time monitoring.

“Agentic AI without governance is a liability,” a Microsoft Azure AI executive stated during the report’s launch. “With Azure, every action an agent takes can be traced, reviewed, and constrained by policies that align with the airline’s operational and legal obligations.”

The report highlights Azure’s integration with Amadeus’s Altéa passenger service system, which runs the back-office of over 130 airlines. By embedding agentic AI into Altéa via Azure, the partners ensure that agents operate with full context of passenger data, flight schedules, and business rules—while inheriting the system’s built-in compliance controls.

Use Cases: Voice Rebooking, Commerce, and Beyond

Voice Rebooking and Customer Service

The most striking example is voice-based rebooking. During irregular operations (IRROPS)—such as storms or air traffic control strikes—call centers are overwhelmed. An agentic AI can handle voice calls, understand passenger intent, rebook flights, and issue vouchers, all within airline policies. The report notes that Azure Cognitive Services combined with Amadeus’s domain expertise enables natural, context-aware conversations that comply with fare rules and ticketing regulations.

Commerce and Dynamic Offer Management

Agentic AI is poised to transform airline commerce. Rather than static rules, AI agents can dynamically construct offers—bundling seats, bags, and meals—tailored to individual travelers in real time. The Amadeus Dynamic Pricing Engine, now infused with Azure AI, allows carriers to test and deploy AI-driven offer optimization with A/B testing and explainability dashboards that satisfy revenue management teams and auditors.

Digital Marketing

Airlines spend billions on digital marketing, yet most campaigns are broad and reactive. Agentic AI can analyze search patterns, loyalty data, and external events (concerts, conferences) to deploy hyper-personalized ads and promotions. Azure’s integration with Microsoft’s advertising platforms ensures that these campaigns respect privacy boundaries and opt-in permissions, the report says.

Aircraft Operations and Maintenance

While the excerpt focuses on commercial workflows, the report extends to operational areas. Agents can monitor aircraft health data in real time, trigger maintenance requests, and order parts—coordinating with ground crews and MRO suppliers. Azure IoT and Digital Twins services provide the foundation for these multi-agent systems, where AI agents reason about complex, time-sensitive logistics without human bottlenecks.

Technical Architecture: Amadeus on Azure

The technical backbone involves Amadeus’s cloud-native platform running on Azure Kubernetes Service (AKS), leveraging Azure OpenAI Service for large language models, and Azure AI Search for retrieval-augmented generation (RAG). This allows agents to ground their reasoning in airline-specific data—flight schedules, passenger records, maintenance logs—while maintaining the highest security posture.

Crucially, the report emphasizes that agents are not monolithic. They are composed as modular “skills” orchestrated by a meta-agent that routes tasks to specialized sub-agents. This architecture improves scalability, simplifies auditing, and isolates failures. A skill for rebooking cannot accidentally invoke a skill for payment processing without explicit authorization, enforced through Azure API Management and Managed Identities.

Safety, Security, and Responsible AI

Airlines are prime targets for cyberattacks, and agentic AI introduces new vectors. Prompt injection attacks could try to manipulate an agent into unauthorized actions. Microsoft’s Prompt Shields, integrated directly into the Azure AI stack, automatically detect and block such attempts. Moreover, all AI outputs are filtered through Azure AI Content Safety to prevent harmful or brand-damaging responses.

The report dedicates a full chapter to Responsible AI. Amadeus and Microsoft have adopted a multilayered approach: fairness testing on booking outcomes, transparency through model cards and visual explainability tools in Azure Machine Learning, and human-in-the-loop overrides for high-stakes decisions. Airline operations controllers can approve or reject agent-driven actions via a unified dashboard.

Industry Reception and Challenges

Early adopters are reportedly enthusiastic. A major European carrier is piloting the voice rebooking agent, seeing a 40% reduction in average handling time during disruptions. A North American low-cost carrier is testing the commerce agents, with initial A/B tests showing a 7% uplift in ancillary revenue. But challenges remain.

Change management is daunting. Airlines have deeply ingrained manual processes and unionized workforces. The report advises a phased rollout, starting with supervised autonomy where agents recommend actions for human approval before graduating to full autonomy in well-defined, low-risk scenarios.

Data quality is another hurdle. Agentic AI is only as good as the data it grounds on. Amadeus and Microsoft are investing in data fabric solutions on Azure to unify fragmented sources—PSS, loyalty, finance, operational databases—into a cohesive semantic layer, reducing hallucinations and improving decision accuracy.

What’s Next: From Single Agents to Multi-Agent Ecosystems

The report forecasts that by 2028, airlines will move from isolated agents to multi-agent ecosystems where dozens of specialized agents collaborate. A rebooking agent will negotiate with a catering agent to ensure meal preferences are honored on a new flight; a marketing agent will coordinate with a loyalty agent to offer targeted compensation during delays. Azure’s multi-agent framework, currently in preview with Amadeus, enables such orchestration with decentralized, event-driven communication.

Microsoft and Amadeus are also exploring how agents can cross enterprise boundaries—for example, an airline agent communicating with an airport agent or a hotel chain agent—using Azure’s trusted execution environments and decentralized identity protocols. This vision of an “AI-powered travel ecosystem” could fundamentally reshape how the industry collaborates.

Conclusion

The Amadeus-Microsoft report draws a clear line in the sand: agentic AI is no longer experimental for airlines. With Azure’s governance and security, the technology is ready for production, promising to transform customer experience, operational efficiency, and revenue management. The next two years will be critical as early adopters set the standard for responsible, scalable deployment. For airlines that get it right, the payoff could be a more resilient, responsive, and profitable business model.