AI Is Reshaping Barcelona's Business Scene — But the Risks Are Keeping Owners Up at Night

Nearly 40 percent of small and medium-sized businesses in Catalonia have either piloted or fully deployed some form of artificial intelligence tool in the past eighteen months, according to figures published in June by the Barcelona Chamber of Commerce. The number sounds optimistic. The reality on the ground is considerably more complicated.

Across the city, from the co-working floors of the Torre Glòries technology district to the family-run wholesale shops on Carrer del Consell de Cent, the conversation about AI has shifted. It is no longer about whether to adopt it. It is about who bears the cost when things go wrong — and who gets to decide what wrong even looks like.

Promise and Pressure in Poblenou

The neighbourhood of Poblenou, home to the 22@ innovation district, has become the most visible laboratory for this tension. Startups there are selling AI-powered customer service tools, demand forecasting software, and automated legal-contract review to Barcelona's mid-sized firms. One platform, developed at the incubator at Carrer de Pallars 193, claims it can cut administrative overhead by 30 percent for retailers with between ten and fifty employees. Several shops on the Passeig de Gràcia have signed up. The pitch works. The complications arrive later.

Three restaurant groups operating in the Gothic Quarter told The Daily Barcelona they abandoned AI-driven scheduling tools within six months of deployment. The software consistently underestimated demand during local festivals and over-staffed on slow Tuesday afternoons, producing wage bills that ate into the margins the technology was supposed to protect. None wanted their names published, citing ongoing supplier relationships. The pattern, however, matches complaints filed with the Agència Catalana de Consum in the first quarter of 2026, which logged a 22 percent year-on-year increase in disputes involving automated business software.

Then there is the question of data. AI tools require feeding. In practice, that means uploading customer records, purchase histories, employee schedules, and sometimes payroll data to servers that may sit outside the European Union. The General Data Protection Regulation theoretically covers this, but enforcement is slow and fines are rarely fast enough to matter. The Barcelona Data Commons, a civic initiative operating out of the Canòdrom digital rights centre in Sant Andreu, has been running workshops since February warning local business owners about consent clauses buried in vendor contracts. Attendance has tripled since April.

Ethics Without a Rulebook

The ethical dimension is where many businesses feel most exposed. AI hiring tools, for example, are being marketed to Barcelona's logistics sector — a major employer given the city's port handles roughly 3.5 million containers a year. Algorithmic screening can process hundreds of CVs in minutes. It can also encode the biases baked into historical hiring data. The Fundació Factor Humà, a Barcelona-based human resources association, published a report in May flagging that several Catalan firms had adopted screening tools without conducting any bias audit, partly because no Catalan regulation yet mandates one.

The European AI Act, which began phasing into force earlier this year, does establish risk categories and compliance timelines. High-risk applications — including employment screening and credit scoring — face the strictest rules, with full obligations kicking in for most operators by August 2026. But awareness among smaller firms is patchy. A survey conducted by the UPF Barcelona School of Management in May found that 61 percent of SME owners in the metropolitan area had either not heard of the AI Act or could not describe a single obligation it imposed on them.

For businesses trying to navigate this sensibly, the practical advice from compliance specialists is blunt: demand a data-processing agreement before signing any AI vendor contract, ask explicitly where model training data originated, and run a basic impact assessment before deploying anything that touches employee or customer decisions. The Barcelona Activa agency, the city's official business development body, has a free AI readiness programme running through September — details at its offices on Carrer de Llacuna in Poblenou. Signing up takes twenty minutes. Undoing a bad AI decision can take considerably longer.