Alicante shows that useful innovation is now born tied to a sector: health, the environment and data sovereignty

There are awards ceremonies that amount to little more than a photo op, and others that, read carefully, anticipate trends. The 32nd edition of Fundeun's New Business Ideas Awards, held at Santa Bárbara Castle, belongs to the second group. Of the 85 entries submitted, the jury selected 50 finalists and awarded 15 projects with prizes of up to 4,500 euros, while also recognizing the foundation's former president, Antonio Fernández Gómez, with the Business Knowledge distinction. Beyond the event itself, what is interesting is the map of priorities drawn by the body of proposals as a whole.

Three of the winning projects connect directly with the most current conversation about artificial intelligence. Atis, an assistant trained on regulations and technical documentation for Social Work, is almost a textbook example of what is taking hold in the sector: the vertical domain agent. As opposed to the generalist model that knows everything only halfway, the tool that deeply understands a regulated field is gaining ground and therefore offers greater precision and a better regulatory fit. In healthcare or social services, that specialization is not a luxury, it is a requirement.

The second name to highlight is NeuroBlock, aimed at enabling companies and organizations to create, train and deploy their own models while retaining control over their data. Here the innovation lies not so much in the algorithm as in the answer to an underlying concern: technological sovereignty. Within the framework of the AI Act and the debate over dependence on foreign cloud providers, the ability of an SME or a public administration to train models without handing its information over to third parties goes from being a technical option to a strategic decision. Reef AI, which applies AI and citizen science to reef monitoring, and TuyTu, focused on protecting personal images against deepfakes, complete a trio that shows AI as a tool for concrete problems, not as an end in itself.

It is worth not losing sight of the rest of the list of winners, because that is where the real context lies. Early diagnosis through biomarkers in tears (LacriDx), rapid detection of Listeria with CRISPR technology (L-Track), bactericidal surfaces for healthcare settings (Zenithia) and several agri-tech proposals show that the innovation ecosystem is not reduced to AI: it uses it where it adds value and dispenses with it where it is not needed. That maturity is, perhaps, the best news.

The editorial reading is optimistic and sober at the same time. Competitions like Fundeun's serve a discreet but valuable function: they turn ideas into viable business plans thanks to specialized mentoring and, along the way, teach us what kind of innovation is germinating far from the major headlines. If the signal Alicante is sending is representative, the immediate future of applied AI runs through the vertical, the regulated and the sovereign. Three rather unflashy words and, precisely for that reason, promising.