The travel industry keeps talking about AI agents that will plan and book trips for customers. But inside a real travel company, the more urgent AI opportunity may be less visible: emails, Excel files, reconciliations, first-level support, API integrations and post-booking operations.
That was the clearest message from my conversation with Agis Anastasiou, Vice President at Ask2Travel and Managing Director at AheadRM, the company behind the Entrada travel reservation platform. Agis sits in a useful position. He sees AI from both sides: as a technology operator building travel software, and as a travel executive working across agency, consolidation, corporate, marine, leisure and VIP travel operations.
His view is practical. AI is already creating value, but in travel agencies, the first serious adoption wave may not be autonomous customer booking. It may be workflow cleanup.
“Wherever you do manual repeated tasks, this equals some kind of AI solution,” he told me.
That sentence is a useful way to understand where AI is entering travel companies now. And the direction is not any different than for SMEs in other fields.
AI is already changing the software factory
On the technology side, AI is no longer theoretical for AheadRM. Agis says his development team is already using Claude for refactoring, migrating old code to new technologies, supporting new API integrations and reducing manual coding work. The tool is not replacing the development team. It is not building full travel technology projects alone. But it is changing how much work developers can get through.
“When needing to integrate new APIs, it’s not that it can handle it solely without any human coding,” Agis said. “But it reduces the manual coding time dramatically.”
The measurable impact is significant. Agis estimates that some development work is now around 30% to 40% faster. That does not mean immediate headcount reduction. In travel technology, a developer who also understands the travel industry is not easy to replace. The value is not only coding ability. It is domain knowledge: suppliers, reservations, ticketing, mapping, workflows, rules and operational edge cases.
“When you have IT-skilled people that also know about the travel industry, you value this combination” he said.
The bigger impact is on the product pipeline. Travel technology companies usually have more roadmap than delivery capacity. API integrations, legacy migrations, platform improvements and customer requests all compete for developer time. If AI makes development faster, the company does not simply do the same work cheaper. It can attempt more work.
“It makes us more ambitious,” Agis said. “It makes us want to engage in much more projects that probably in the past we would need to pipeline in a period of two, three years.”
Then he used a phrase that will sound familiar to anyone who works with travel technology vendors: “Emptying the pipeline is very helpful.”
That is an important point. On the technology side, AI is not only a cost-saving layer. It is becoming a roadmap acceleration tool.
Agency operations are a harder problem
The agency side is different. Ask2Travel still has workflows that remain manual. Agis mentioned flight ticketing, post-booking support, hotel confirmations, operational cases, reporting and reconciliation. He separates the digitalization opportunity into two broad area: travel design and post-booking operations. Both can benefit from AI, but both are more complex than they look from the outside.
“Most of the services we offer consist of components coming from live systems where the pricing is real time,” Agis said. “There I think is where the complexity and the difficulty begins.”
That complexity is why he does not see AI simply replacing travel consultants. For simple cases, AI can help with first-level operations. It can prepare information, classify requests, support reconciliation, or assist an agent before a human takes over. But for complex operational cases, especially where escalation is needed, he still wants a person involved.
“It’s not one category specifically that’s a no-go,” he said. “But in every category, the more complex and demanding cases, there you want a human to handle them.”
That is the more realistic model for AI in agencies: not AI instead of agents, but AI before the agent, around the agent and underneath the agent.
When I asked Agis which workflows he would automate first if the perfect AI solution existed today, he did not start with a consumer chatbot. His first answers were operational: first-level support cases, reconciliation tasks across accounting and product departments, and the trip planning.
That order matters. The first two are internal. They do not create a flashy demo. But they may create the fastest business case. In many agencies, Excel is where unofficial systems live. Reports, reconciliations, supplier checks, margin calculations, RFP inputs and ad hoc customer information often sit outside structured workflows. If AI adoption starts anywhere, this is one obvious place. This is where Agis sees immediate AI value.
“The first two things I would do is connect my AI tool to my email and map whatever I do to Excel,” he said. “This is also the process we’re going through internally, trying to identify each workflow we have, what’s done repeatedly and manually, to transition to an AI agent.”
Agis is blunt about the spreadsheet problem.
“Whenever I hear somebody storing Excel files somewhere and maintaining information in Excel files, it’s mad,” he said. “I’m in a task to eliminate Excel files.”
Adoption is workflow change
One of the strongest points from the conversation was that travel agencies do not only face a tooling problem. They face an adoption problem. Many agencies are still heavily dependent on email, phone calls and WhatsApp. Many smaller agencies do not have an AI strategy. Some are waiting for larger technology providers, GDSs or platform partners to deliver something easy to adopt. The first barrier is basic: many companies do not know where to start.
The second comes immediately after: which AI tool should they use, for which workflow, at what cost, and with what level of trust?
But the deeper issue is workflow change. Agis made a useful distinction. AI adoption is not simply putting an AI layer on top of the way the company already works.
“When you’re talking about AI strategy, you’re not just talking about putting an AI agent on top of what you’ve been doing until now,” he said. “It means changing and transitioning the way we’ve been doing things till now across all your departments where AI is applicable.”
That may be the most important line in the interview. If a travel agency has fragmented workflows, AI does not automatically create operational discipline. If request handling depends on individual habits, inbox logic, spreadsheet workarounds and informal knowledge, AI may expose that fragmentation rather than solve it.
This also changes how success should be measured. For Ask2Travel, Agis would look at whether AI increases basket value, reduces time spent by consultants and handles simpler requests without unnecessary human work. But the more interesting test is what happens when the AI does not fully solve the case. This matters because many travel cases will still be escalated. AI can still create value if it gives the human agent a better starting point.
“Even if something is escalated to a human person, I’m interested in the quality of information you get regarding the request,” Agis said. “Is it much more prepared for the human to handle it? Does it reduce a lot of search time from the human?”
That is a better way to think about AI in travel operations. Success is not only full automation. Sometimes success is a better-prepared human.
Will AI strengthen or weaken travel agencies?
The common fear is that AI will weaken travel agencies by giving travelers more tools to search, plan and book directly. Agis is more optimistic.
“I’m optimistic that it will strengthen them,” he said.
His reasoning is simple: travel looks easy at the interface layer, but remains complex behind the scenes. Search, booking, servicing, changes, supplier communication, pricing, support and exceptions still require operational knowledge. That does not mean every agency benefits automatically. Agencies that continue working the same way may fall behind. Customers will expect faster service, better tools and more intelligent support. Agents will need systems that help them handle more complexity with less manual work.
As Agis put it:
“Even to your existing customers, they will expect from you to be able to offer them new tools in order to do the business. If you think you will be supporting them and facilitating in the same way you’re doing today — question mark.”
The risk for travel agencies is not that AI instantly replaces them, rather that AI changes customer expectations while many agencies are still organized around old workflows: email, phone, WhatsApp, Excel and manual follow-up.
In travel, AI adoption often starts behind the scenes. That makes it less flashy. It does not make it less important.