AI for travel agencies: Where it saves hours and where human agents remain essential
The most concrete application of artificial intelligence in travel agencies today is not a chatbot for travelers. AI shows its greatest potential in tasks that take agents hours every day: processing inquiries, creating offers, searching for suppliers, calculating costs, and operational administration.
For DMCs and tour operators working on individual and tailor-made trips, this may matter far more than another text-generation tool. A complex inquiry sets off a series of tasks: reading the email, entering details, checking availability, comparing prices, building an itinerary, calculating margins and communicating with several suppliers.
AI is now beginning to handle precisely this part of the process.
From the email to the first version of the offer
Croatian Lemax is one of the regional examples of development in this direction, but there are also many others. Its AI Sales Assistant can analyze incoming inquiries and automatically prepare an itinerary within the Lemax system. Some agencies, such as Nomag, are also developing AI automation for inquiries and other business processes themselves.
This changes the logic of the sales agent's work. Instead of first entering the dates, number of passengers, destinations, and other requirements, he or she gets the initial structure of the offer, which he or she then checks and adjusts.
Greece’s Ten06 takes a similar approach with its Atlon system. The platform uses incoming inquiries to identify dates, passenger numbers, budgets and special requests, then connects that information with supplier systems. It prepares a draft itinerary for the agent to review before sending it to the client.
According to Ten06 itself, the process of creating a complex offer that can take up to four to six hours manually in certain cases can be reduced to approximately 20 to 30 minutes of review and refinement. This is a result that the technology provider points out, and it should be viewed in that context. However, it shows where the greatest return from AI is currently sought: not in replacing agents, but in eliminating several hours of repetitive work.
AI is no longer limited to sales
An even more interesting application appears in the operation.
Greek travel-tech company Cyberlogic presented Stop Sales AI Automation in March 2026. The system recognizes stop-sale requests in incoming messages, extracts the hotel, dates, and conditions, and sends structured data to the business system. This is especially important for tour operators and agencies that otherwise work manually to update stopsales.
This is a good example of a process that the traveler will never see, but can have a direct business impact. Hand-processing stop-sale messages takes time, and late entry increases the risk of errors or the sale of capacity that is no longer available.
AI thus becomes interesting precisely in places where tourist companies have a large volume of small administrative tasks.
Jyper applies a similar approach to quotes, calculations and supplier contracts. On its website, the company cites Liberty Adriatica as an example: during onboarding, the system automatically matched 90 percent of existing contracts to their suppliers. Jyper also shares a customer account that automated calculations saved around 20 working hours a week. These are customer claims published by the technology provider, not an independently measured benchmark.
The real ROI is not the number of AI tools
For agencies and tour operators from the region, the question is not whether to use AI, but which specific process should be automated.
If ten employees manually transfer data between email, Excel, the booking system, and the PDF quote every day, there may be more room for savings here than in a generic chatbot on the website.
The simplest test can start with a few internal indicators: how much time passes from the inquiry to the first offer, how many work hours are spent on a single complex offer, how many offers an agent can process per day, and how many manual interventions are required after confirmation. Only then does it make sense to compare the technology.
There is also another requirement. Automation is only useful if AI has access to quality data. A flawed pricing policy, an outdated contract, or a poorly structured database will not become more efficient just because AI processes it. On the other hand, integrations with booking systems, CRM, suppliers, and internal databases become just as important as the AI model itself.
What does it actually remain for the agent?
The boundary is quite clear. AI can read the query, structure the information, find options, calculate the price, and prepare the first version of the program. It can also warn about missing data or an operational problem.
But the decision of whether a particular hotel is really right for a specific guest is not just a matter of data.
The same applies to negotiations with suppliers, complex group requests, crisis situations, VIP guests, or programs where local experience makes the difference between a good and excellent trip.
That’s exactly why the most interesting models that appear in travel tech today generally don’t try to take the human out of the process. The agent becomes a point of control, but spends less time compiling offers and entering data.
This is especially important at a time when the owners and directors of travel agencies are facing increasing problems finding quality workers and the cost of labor. Generic AI tools can also be used to automate and improve business processes; these tools can now be very cheaply used to shorten the time it takes to complete tasks. Perhaps the problem is to change the mindset of workers and explain that things have to be accelerated today while maintaining the existing quality; that is a different problem now.,
For DMC's and tour operators, this could be a more significant shift than the AI technology itself. If an agent with the same working hours can process more quality inquiries, respond faster, and spend more time talking with a client or supplier, automation begins to have very simple business calculations.
