Challenges
Uniline, one of the leading travel agencies in the Republic of Croatia, faced challenges in processing numerous queries for group trips from various markets (B2B segment), as well as queries for creating offers for skiing. A large number of queries that are sent in the free form of text, as well as the complexity of the collected information, and the available time for submitting bids often result in the need for overtime work, often on weekends, all in order to maintain the level of quality of service, achieve sales and ensure business continuity. In addition, the traditional way of working relies on manual data processing, which puts pressure on employees and increases the possibility of errors.
Given the growing number of inquiries and customer expectations, it has become clear that the process needs to be modernized. Uniline has identified digitalisation and automation as key steps to increase efficiency, reduce costs and provide better customer service. The aim was to develop a system that would speed up the processing of inquiries, identify inquiries from major clients and enable the automatic generation of standardized offers and improve the overall customer experience.
Solutions
In order to address the challenges, EDIH Adria proposed two concrete solutions:
Prototype of the system for ranking queries via a structured form: This LLM prototype uses the OpenAI model GPT-4o, located on an OpenAI infrastructure that allows real-time processing and analysis via APIs. The system ranks queries precisely on the basis of input data from queries sent to the agency via e-mail, as well as on the basis of predefined criteria. The established criteria used by the prototype are: priority agents, the value of the query in terms of the quantity and value of the services requested, the term of the trip taking into account the criteria that certain time periods have priority, the keywords in the title of the query. The graphical interface is realized through Streamlit software package that serves to enter data through a structured form, which simulates the receipt of e-mail messages and also serves to display ranking results with detailed information on the logic based on which the rank query was determined.
Video: UNILINE_ranking of group travel queries_kina
Prototype of the system for making offers for skiing: This solution uses the LLM system to process unstructured queries received via e-mail or chat and make informal offers and recommendations to the customer. The prototype was developed with a web interface to enter queries in the form of free text, allowing users to make queries using natural language. Based on this information, the system searches for data from available sources, such as Uniline Group websites, as well as internal databases, in order to generate customized offers for customers.
Video: UNILINE_Crispo_supply of skiing arrangements
Both prototypes are flexible and customizable, designed with a simple and intuitive web interface, which allows easy testing and visualization of possible integrations with existing tools and systems in the Uniline Group's business.
Results and Benefits
The test phase of the solution has shown the potential for significant benefits of introducing digitalisation of existing processes:
- Faster answers to queries: Automation can make it possible to significantly reduce the time it takes to process a query, which would provide customers with the ability to react quickly.
- Better tailored offers: A combination of publicly available and internal data could make offers more precise and relevant.
- Increased productivity: Uniline Group employees could devote more time to more complex tasks, instead of manually searching for newly received emails and routine data entry.
- Wider availability: The systems support multiple languages, enabling better communication with international clients, removing language barriers.
- Standardisation of communication: It is possible to define a standard set of data that would be sent to each query, which in further stages of development enables the creation of an API through which Uniline Group partners could directly connect to databases and accelerate the interaction even more, which would make the Uniline Group's offer more responsive and competitive.
Lessons learned
The project emphasized the importance of structured data and quality preparation for the successful implementation of digital solutions. Clearly defined processes and regular database updates have proven to be key to maintaining the high precision of the system. Involvement of employees in the development and testing of the system was crucial for a better understanding of the needs and ensuring the adaptability of solutions to their daily work.
In addition to technical aspects, the importance of continuous cooperation with the implementation team and flexibility in adapting solutions to feedback was also learned.
Conclusion
With the implementation of such solutions, Uniline Group will not only optimize its business processes, but also lay the foundations for a wider application of artificial intelligence in tourism. Systems enable increased efficiency, reduced costs, faster response to queries and improved user experience.
TBI Uniline project shows how modern technology can facilitate business and bring measurable benefits, serving as an example of good practice and potential for similar implementations in other sectors. By introducing similar solutions, Uniline Group will be better prepared for future challenges and additional growth, while maintaining a high level of service quality.
English 
Croatian 
Italian