Can AI Take Over Travel Tasks?

AI is transforming corporate travel workflows, with AI agents now taking on work once performed by humans: booking trips, applying travel policies, processing post-booking changes and pumping out program intelligence.  

Generative AI adoption is increasing in corporate travel, but agentic AI is in its early stages. In a McKinsey–Skift survey conducted in summer 2025, 90 percent of travel executives said their organizations use generative AI in some capacity, yet 38 percent said they are not using agentic AI at all. While 22 percent reported broad use of generative AI across their organizations, only 2 percent said the same of agentic systems. 

Agentic vs. Generative AI 

Agentic AI refers to tools that can plan and execute tasks on a user’s behalf—booking trips, modifying itineraries, applying policy, or resolving itinerary disruptions. Unlike generative AI, which performs a single task related to a user prompt, agentic AI can make decisions and take multiple actions to complete a complex, multi-step task. 

Tauseef Hussain, who leads the data department at Areka Consulting, broke it down like this: 

• Machine learning draws on past data to identify patterns and make predictions.  
• Generative AI produces new outputs.  
• Agentic systems use generative models to interpret prompts, then go a step further by completing tasks. 

Agentic systems might connect to policy engines, booking tools or other internal systems. Some also incorporate machine learning to refine results over time. 

Use Cases  

Booking and applying policies: “Agentic AI broadly is just software taking actions based on inputs or a problem statement,” said Atyab Bhatti, CEO and co-founder of SkyLink, a booking platform with agentic AI capabilities.  

SkyLink has recevied lots of attention over the past two years, winning the BTN Group’s Innovation Faceoff competition in 2024 and since then gaining clients like consulting giant McKinsey & Co. and entertainment company Sony. Additionally, travel management companies like BCD Travel and Amex GBT have launched the technology.  

As a booking tool it integrates with content sources at the agency, whether that’s global distribution systems or other content streams. It takes corporate policy and traveler profile and preferences into consideration as it looks to return itinerary options for searches, with a balance that is defined by the corporate. While other tools can do this, SkyLink goes further with large language model responses that explain policy issues and suggest alternatives, much as a human agent would do with live booking advice. And when the traveler is ready to make the booking, the technology can draw the information from the traveler’s GDS profile, use the designated form of payment and the traveler does not need to make the keystrokes or take the time to make that happen.  

For the agency, that’s a time saver and a money saver—particularly in an era when travel agent human resources can be scarce. But it’s also an efficiency and satisfaction driver for the traveler and the corporate, with compliance to policies and payment mechanisms baked in.  

There’s plenty more upside for agentic AI when it comes to booking. Integration with calendars to recognize meetings and take action on employee intent is on its way. SkyLink is just one example of a technology laying the foundations now.  

Post-booking changes and disruptions: Agentic AI can be used to streamline post-booking tasks that previously consumed significant agent time, including exchanges, cancellations and travel disruptions. Agentic tools from companies like Oversee and Acai Travel automate much of this workflow by analyzing fare rules, travel policy and available rebooking options. They can suggest alternatives and process changes and pass tasks on to human agents when they cannot complete them autonomously. 

Ron Glickman, head of sales at Acai Travel, said agentic AI can step in during disruptions to assess refund eligibility, compare rebooking options across carriers, and suggest alternatives that still allow the traveler to get to their meeting on time. Agentic tools can also review fare rules and corporate policy to determine whether changes are permitted and what costs apply, an Oversee spokesperson added. 

Glickman said agentic AI is better positioned than humans to respond instantly during high-stakes disruptions, because it can detect, decide and act before a human even knows there’s an issue. 

Expense reconciliation: Hussain said agentic AI improves how travel bookings are matched with expense records, raising automated reconciliation accuracy from about 65 percent with legacy tools to as high as 90 percent with agentic systems. He added that separate AI agents review different parts of the trip, such as flights and hotels, and compare notes to make sure the details line up correctly. 

AI as a Full-Time Travel Management Equivalent: Data Analytics provider PredictX this year launched AI workspace, Cogent, designed to act as an “intelligent workforce of AI agents.” 

The workspace is powered by data in the PredictX platform, including data pulled in from travel management companies, expense reporting systems, corporate cards, online booking tools, global distribution systems, suppliers, HR systems and a company’s general ledger.  

With that, users can not only make natural language inquiries about their data—a use of generative AI—but agents in the workspace can autonomously monitor spending, contracts and policies to find savings opportunities and recommend actions to travel buyers. 

For example, an AI agent can monitor key performance indicators and, knowing a program’s goals, look for savings opportunities. A vendor performance management agent can analyze a supplier relationship to provide advice to a buyer before a negotiation, and a contract management agent tool can analyze a contract and point out comparisons between an old and new version of a contract. 

How long might it take to ramp that process to a point where a supplier AI agent and a buyer AI agent are negotiating with each other? New industry technologists are building toward that goal.  

How quickly it will happen and to what extent companies will adopt technologies that make actual decisions without human intervention will hinge on company culture and risk appetite, said Acai Travel’s Glickman.  

But, he added, you make that call with trusting humans too.  

“It’s the same as it is with a human agent… how much do you trust that [agent or AI] to follow your policy and use judgment,” he said. Some companies will hold tight the reins on when AI can make decisions, while others will press forward as quickly as possible to more AI autonomy.