What agentic AI can do for you

PRICING

Amit Singh, Fareportal: Today’s travel advisors spend an enormous amount of time comparing options across suppliers, fare classes, loyalty benefits, ancillary fees and client preferences. Agentic AI has the potential to fundamentally change that process. Rather than simply returning search results, an AI agent can help advisors understand the traveler’s priorities — whether that’s lowest cost, best value, preferred airlines, sustainability goals or loyalty status — and continuously evaluate thousands of options against those requirements.

The most powerful opportunity is moving from “best price” to “best fit.” An agent could proactively recommend itineraries that balance cost, convenience and traveler preferences while explaining the trade-offs. This allows advisors to spend less time searching and more time delivering personalized guidance and service to their clients.

Brad Johnson, Sabre: Finding the best offer for a consultant’s client starts with matching traveler intent and data with available offers. An appropriately tasked AI agent can process large and disperse sets of data, such as profiles, preferences, policies, budget and travel history, to support offer recommendations from a broad set of travel itinerary options. 

This significantly reduces the heavy lift on the travel advisor and can better create tailored options more likely to convert. 

In addition, as the travel advisor engages with the client and offers are refined, the AI agent can continuously learn and improve for future requests.

Beyond supporting the up-front travel query, agentic AI can also support monitoring and predicting price changes to surface preferred options as they emerge and even act on them with approval. This shifts pricing from manual search or repeated queries to smart automation and execution. 

Robin Lawther, Expedia Group: I think the real opportunity is helping advisors with both destination and hotel selection, particularly when they’re supporting customers traveling to places they may not be as familiar with themselves. Once a destination has been selected, AI can help identify the “next best” alternatives if, for example, a customer’s preferred hotel is outside their budget. 

There’s also value post-booking. Advisors could set up alerts to monitor pricing and be notified if rates change. So if a client’s first-choice hotel was too expensive at the time of booking, AI could flag a future price reduction and create an opportunity to rebook them into their preferred property.

Gilad Berenstein, Brook Bay Capital: On pricing and fragmented content: These two issues are really two sides of the same coin. The advisor’s job of manually shopping across Edifact, NDC and a dozen hotel aggregation platforms is exactly the kind of tedious, high-volume, low-judgment task that AI agents were built for. This is a swarm problem, not a super [AI] agent problem. 

You don’t need one megabot that understands the entire travel ecosystem; you need specialized bots — one for airfares, one for hotel content, one for pricing logic — that can temporarily connect, compare and surface the best-fit option for that specific client at that specific moment. 

The technology to do this already exists. What’s still maturing is the depth of the integrations and the trust layer that lets these bots reliably talk to legacy systems.

FRAGMENTED CONTENT AGGREGATION

Molly Johnson, Tern: Advisors lose whole evenings to assembly work, one browser tab at a time, and this is exactly where generic AI falls short. 

The best rates are often advisor-specific rates that live behind suppliers’ portals, which general AI tools can’t reach, and any supplier-built AI only shows you that one supplier. 

Solving this takes a platform with direct supplier integrations because the hard half isn’t the search, it’s the normalizing of mismatched data. The same cabin or fare shows up with different inclusions depending on what portal or website you book it on. Tern has already proven that this is possible to fix with direct integrations while protecting the host, agency and advisor commission structure. 

Imagine that in one chat conversation, you can search live options, browse day-by-day itineraries, compare cabins down to square footage and live pricing, pick a booking supplier and apply everything to the trip all in one place. That is how we think this headache will be solved best.

Singh: One of the largest friction points in the travel ecosystem is the fragmentation of travel content across multiple systems and channels. Advisors often need to search and compare information from numerous sources before creating a complete recommendation. 

Agentic AI is well-suited to orchestrating work across these disconnected environments. Instead of requiring an advisor to manually access multiple content repositories, an AI agent could simultaneously search airline, hotel, rail and other supplier content sources, normalize the results and present a unified set of recommendations. The advisor remains in control, but much of the repetitive work of gathering and organizing information can be automated.

Brad Johnson: Content fragmentation is not a new topic, and much has been done to normalize data sources and access to content in an easily consumable way. This is the promise of the Sabre Mosaic Marketplace. 

But as travel advisors are asked to piece together components of the traveler’s journey from more sources — and with broader information sets — the time spent researching and booking the trip can be wildly inefficient. This is where an AI agent can help support the travel consultant to quickly shop, compare and book travel options across Edifact, NDC, low-cost carriers and millions of hotel options in a single conversational flow. 

With well-structured and AI-friendly APIs, a large language model application can readily support the travel advisor’s informational queries alongside shopping and fulfillment actions without trying to pull the trip together from dozens of access points. 

This technology also significantly reduces friction, allowing agencies to access diverse options without managing dozens of point-to-point integrations.

SCHEDULE CHANGE MANAGEMENT

Berenstein: This is maybe the easiest win of the five and also the most obviously overdue. Turning a schedule change into a background process that only escalates to a human when judgment is genuinely required is not a hard AI problem. It’s a workflow and integration problem.

Molly Johnson: Schedule changes are the best case for AI that runs quietly in the background, because most of the work isn’t making changes, it’s deciding which ones matter. A flight that shifts 20 minutes is just noise, but a flight that shifts enough to cause a missed connection threatens the trip. Telling those apart is exactly what agentic AI needs to be good at to be useful.

We’re starting to build this loop at Tern with email triage. It reads each change as it lands, checks it against the itinerary, evaluates severity, and flags the updates and issues to the advisor, all before they open their inbox. Right now, Tern’s AI highlights what needs attention (rather than making changes on its own), so it leaves decision-making to the advisor.

Brad Johnson: Schedule changes are high-impact use cases for agentic AI due to the volume of contributing factors as well as the opportunities to improve the traveler experience. AI can first be leveraged to help sense and predict disruptions, proactively notifying customers of the likelihood for impact due to weather, strikes, equipment delays, geopolitical events and other scenarios and help stay ahead of impact. 

When disruption does occur, agentic AI can be tasked to help support engaging the traveler and automating the search for alternatives, rebooking or canceling travel. This can be interwoven seamlessly with travel advisor oversight when additional support is needed.

For travelers, this means immediate resolution instead of uncertainty or waiting for support. For advisors, there is an opportunity to provide better value to their clients by shifting from reactive servicing to continuous, proactive, intelligent travel management.

COMMISSION RECONCILIATION

Lawther: Commission reconciliation is another area where AI could help reduce manual effort. Advisors often need to work across multiple systems, emails and booking records to reconcile payments. AI has the potential to bring relevant information together automatically, helping advisors spend less time on administration and more time supporting customers and growing their business.

Singh: Commission reconciliation involves tracking payments, validating amounts, identifying discrepancies and following up with suppliers: tasks that are often repetitive and time-intensive.

AI agents can help automate many aspects of this process. An agent could continuously monitor bookings, compare expected commissions against received payments, identify missing or underpaid commissions and generate follow-up actions when discrepancies occur. In many cases, it could even prepare supporting documentation and draft communications for advisor review.

By reducing manual reconciliation work, agencies can improve cash-flow visibility, minimize revenue leakage and free staff to focus on client service and business growth.

Molly Johnson: Reconciliation is pattern-matching, which is exactly what AI is good at. It stayed manual for so long because every supplier statement is different, and rules-based software needed a rule for each one. An AI agent instead can read any format, match lines to the expected commission and flag the exceptions, which is what Tern already does with reconciliation today. 

What we still see up ahead is closing the loop after it. That means a one-click way to invoice the supplier and an AI agent that queues those invoices and follows up on its own. The payoff there isn’t just time saved, it’s faster payouts.

GROUP BOOKING MANAGEMENT

Molly Johnson: Groups are where small misses turn into real penalties. A deadline slips, the room block doesn’t fill, and the agency eats the cost. 

A lot of the plumbing for solving this problem already exists in Tern today. Group trips sync down to their subtrips automatically, workflow automations send deposit reminders and payment-deadline emails without the advisor touching anything, and you can ask the AI to summarize every sub-trip in a table or dig through dozens of traveler form responses in one pass.

What agentic AI can still add is the watching: tracking sign-up pace against the contract and telling the advisor it’s time to release rooms three weeks before the penalty or reminding you to nudge interested travelers. A forecast you can still act on is the difference between a task on your calendar and an outcome.

Berenstein: The AI-supported request for proposal and group space is genuinely one of the busiest corners of travel tech right now. Monitoring attrition dates, nudging unbooked travelers and dynamically readjusting inventory with a hotel is a perfect use case for an AI agent working continuously in the background, doing the kind of relentless, detail-oriented follow-up that no human wants to do at scale but that directly protects an advisor’s margin.

Brad Johnson: Group bookings can drive significant manual labor associated with tracking attrition clocks, chasing unbooked stragglers and negotiating inventory back with hotels before penalties hit. These processes are typically managed via spreadsheets, emails and phone calls. Agentic AI excels at processing large amounts of distributed data as well as automating tasks that would typically require time-consuming work.

For something like a destination wedding or corporate event, an AI agent could monitor booking deadlines, automatically nudge travelers who haven’t booked and engage directly with the hotel’s system to safely adjust remaining inventory. Instead of just flagging a risk, it can proactively work to resolve it.

Operating within secure guardrails, agents are able to handle the downstream logistics end to end, from communications to tracking and processing inventory changes without constant intervention. The result is less administration, lower risk of error and a shift from chasing details to supporting the group’s experience.

Is an AI agent the best solution for your agency?

Mike Coletta, senior manager of research and innovation at Phocuswright, said agentic AI “could help with just about anything” — but so could other tools. 

“The magic of agentic AI is in the AI agent’s ability to figure things out, make decisions and work autonomously toward goals,” he said. “But it can only be as good as its configuration and its access to data.”

To determine if a problem is best solved using an agentic solution, Coletta suggested a simple diagnostic approach: Ask the right questions first, and let the answers dictate the technology. 

For instance, when considering pricing as a pain point, could an agentic solution be applied to enable travel advisors to find the best price option for their client, whether it’s for the lowest price or the best-fit price? 

Coletta suggested asking these questions: 

• What constitutes “best fit” for a client, and can that be defined clearly?
• Which systems is the AI agent allowed to access, and what are the capabilities those systems provide?
• How thoughtfully and carefully is the AI agent instructed, and what kind of guardrails are in place?
• Is an AI agent exercising judgment really needed for a particular scenario or is a rule-based system sufficient?

If the answers clearly point to an agentic solution, it could be the right fit, Coletta said. But if they don’t, look to another kind of technology.