The ALIA Forum served as a platform to demonstrate how talent and experience can provide solutions for businesses. The event, promoted by the Alicante Digital Intelligence Center (CENID), was sponsored by Lynx View and collaborated with CEV Alicante, as well as the newspapers Levante-EMV, INFORMACIÓN, and Mediterráneo.
Under the theme "From Models to Solutions," five initiatives developed with artificial intelligence were presented. These applications address challenges in cybersecurity, tourism, and the promotion of the Valencian language, adding to a dozen existing applications already serving society.
During the session held at the Science Park of the University of Alicante, project leaders conducted practical demonstrations. One of the presented cases was "Aitana Hotel Assist," a documentary assistant for hotel reception and service, developed by Guillermo Villena and Ernesto Estevanell (both from CENID). This system uses the company's corpus to offer standardized recommendations aligned with the establishment's values, improving consistency, internal policy compliance, traceability, and efficiency.
Two notable cases focused on the use of the Valencian language, considered key for training AI models in minority languages. Daniel Compaño, from Recursos en la Red (RENR), presented the AI editorial translation platform Aitana, developed with the newspaper Levante-EMV. This system, requiring human supervision ('Journalist in the loop'), enhances workflow and linguistic consistency, and is scalable to other languages.
Another application is VeuVal, a voice recognition model (TTS) developed entirely in Alicante by Jose Ignacio Abren (UA). This model, using Style TTS architecture and the Kit ALIA Valenciano corpus, demonstrates that voice interfaces and chatbots with local dialects generate more valuable content. 89,000 phrases from the Valencian Courts (Corts Valencianes) were used for its training.
In the tourism sector, the company You Name it presented "GobernantIA," a tool for managing hotel room cleaning. Aimed at head housekeepers, it considers over 50 parameters, including working hours and workload, addressing demands from "kellys" (hotel cleaning staff). The model uses objective and subjective variables, with an optimization engine and real-time backend.
Finally, Kyndryl introduced "Aitana Fraud Guard," a cybersecurity solution for detecting fraudulent emails and messages ('phishing'). Developed by Francisco Mateu and Pol Torres, this AI shifts from a reactive to a preventive approach, using language models to classify, detect, and explain fraud. It includes a plugin for Outlook and an Android application, with specialization allowing execution in controlled environments.
