Most SMEs do not need to "do AI" in the abstract: they need to fix a specific bottleneck (reading invoices, triaging support tickets, searching internal documentation) and find out whether automating it is worth it. That is why we always start with a diagnostic, not a ready-made proposal.
In Barcelona we work mostly with data and processes in Catalan and Spanish, which shapes which models and tools make sense to use. Not every process is a good candidate for AI: often a classic automation (rules, integrations, scripts) is cheaper, more reliable and easier to maintain.
Part of our job is telling you when AI is NOT the answer, and what risks you take on if you implement it anyway: model errors, dependence on an external provider, per-use costs, and the need for human oversight.