To translate artificial intelligence (AI) strategy into reality, leaders must focus on two operational priorities: Establish custom evaluations (evals) and strengthen data governance.
Establish custom evals
An AI eval is a structured process used to systematically test and measure an AI system's quality, performance and reliability, assessing whether it meets goals like accuracy, safety and usefulness.
Generic evals and benchmarks are insufficient—custom evals are the only way to truly understand an AI's readiness for deployment.
Every travel company must create evals that measure AI performance against exceptional human employees on specific, high-value tasks. This could be complex itinerary planning, real-time disruption management or crafting personalized marketing campaigns.
Examples include:
- Can this agent rebook a canceled flight with 99% accuracy regarding airline interline agreements?
- Does this agent correctly apply EU261 compensation rules across 100 disruption scenarios with 98% accuracy?
- Does the recommendation engine surface properties matching stated guest preferences (e.g., pet-friendly, pool, accessibility) with 95% accuracy?
- Does the revenue management AI match or exceed a senior analyst's pricing decisions across 500 historical demand scenarios?
- Does the duty of care agent locate and contact all travelers in an affected region within the contractual SLA?
- Does the AI correctly identify when to escalate emotionally charged interactions to human agents 98% of the time?
Prioritize clean data and governance
Data governance has shifted from a back-office function to a strategic imperative.
The problem is foundational.
AI systems are only as reliable as the data that feeds them, and probabilistic agentic systems that act autonomously amplify the consequences of bad data. Inaccurate inventory, stale pricing or inconsistent customer records become not just operational annoyances but sources of cascading errors when agents are booking, rebooking and managing disruptions.
The companies closing the gap between AI experimentation and AI transformation treat data governance not as an afterthought but as a precondition. They are investing in data lineage, quality monitoring and clear ownership before scaling autonomous systems.
In the agentic era, governance isn't bureaucratic overhead—it's the trust layer that determines whether an AI can act with confidence or is flying blind.
Phocuswright’s Travel Innovation and Technology Trends 2026
As AI reshapes travel, Phocuswright’s recent report looks at how generative and agentic AI is impacting travel layer by layer.