What AI for Good means to us: Elna Health at MWC26

At 4YFN during MWC26 in Barcelona, we've pitched Elna Health at the AI for Good Innovation Factory Catalonia – AI for Health, where we've been selected as one of five finalists from the Catalan AI and healthcare ecosystem.

We didn't win. emily.AI did, and congratulations to the team. Their respiratory care solution is tackling a very real clinical problem, and we were genuinely happy to see another Catalan health-tech company represent the region on the global AI for Good stage. The value of the experience went well beyond the result, though: it made us think harder about what AI for Good actually means, and what it should mean for a company like Elna.

AI for Good is more than putting AI on a good problem

The AI for Good initiative is the United Nations' leading platform for exploring how AI can help solve global challenges. Led by the International Telecommunication Union, it convenes startups, researchers, governments and other organizations around applications of AI tied to the UN Sustainable Development Goals. The Innovation Factory takes that mission into the startup world, connecting founders with investors, experts and partners to help them scale their impact.

The Catalonia AI for Health chapter brought that perspective to Barcelona. Five Catalan startups, Elna among them, were invited to pitch AI solutions to healthcare challenges at an event organized with the Government of Catalonia and coordinated by the Digital Catalonia Alliance (DCA-IA) and CIDAI.

That context matters, because we don't believe that "AI for healthcare" automatically means "AI for good." Healthcare is high-stakes. An AI system can be technically impressive and still be poorly suited to clinical practice: fast without being trustworthy, plausible without providing the evidence needed to verify it. So the real question is less what AI can do in healthcare than what we should ask it to do, and how we make sure it does that responsibly.

Why Elna cares about AI for Good

We started Elna because of a fundamental problem: medical knowledge is advancing faster than healthcare systems can reliably put it into practice. There is an extraordinary amount of clinical knowledge available, from research and guidelines to institutional protocols and expert experience, but access to it isn't the same as being able to use it well. Clinical teams need to know what is relevant, what is current, and what their own institution has approved; hospitals, meanwhile, have little visibility into what knowledge is actually guiding care. When that knowledge is fragmented or outdated, the consequences are real: variation in care, inefficient processes, preventable errors, and cognitive load on already stretched clinicians.

That's the problem we are trying to solve. Elna is building the infrastructure for trusted clinical AI, the layer that connects evidence, institutions and clinical workflows so that AI in healthcare is traceable, governed and trusted.

It works in three layers. A knowledge engine maps and structures medical knowledge from trusted sources. A governance layer gives hospitals control over their own clinical knowledge, so an institution knows what is guiding care decisions, where it came from and when it was last reviewed. And a clinician-facing app delivers fast, referenced answers grounded in that governed knowledge rather than in a generic model.

We don't want to build an AI that simply gives clinicians another answer. We want one that helps them get to knowledge they can trust and verify.

Responsible AI has to be built in

This is also why we care about the responsible part of AI for Good. Responsible clinical AI isn't a marketing layer added once the product works; it is part of the product. In practice, that means four things.

Traceability. Where does an answer come from? Can a clinician follow it back to trusted medical sources?

Currency. Is the underlying knowledge up to date?

Human oversight. Where should AI assist, and where should qualified healthcare professionals remain firmly in control?

Safety and compliance. How do we build for the realities of GDPR, the EU Medical Device Regulation and the AI Act?

These principles shape how we are building Elna today, and they matter more as generative AI gets more capable. The more convincing AI becomes, the more it matters to know why we should trust it.

Why being a finalist mattered to us

Being selected put our work in front of people well outside our immediate circle, but the most valuable part was engaging with the jury: a panel spanning responsible AI, healthcare systems, health economics, academia and the technology ecosystem, with representatives from the WHO, the Observatory for Ethics in Artificial Intelligence of Catalonia, Tech Barcelona and the University of Barcelona. That kind of audience is useful precisely because it challenges your assumptions.

As founders, it is easy to spend all our time on product, technology, customers and execution. An initiative like AI for Good makes you step back. What problem are we really solving, and who benefits if we succeed? Where could the technology create unintended consequences, and are we measuring the right things? Can this scale beyond the first users and the first market? And most importantly, are we spending our energy where we can create the most meaningful impact?

Those are good questions for any company. For a company building AI for healthcare, they are essential.

AI for Good helps us keep the impact in focus

This is probably the biggest reason we were happy to participate. There is always a temptation to focus on what is technically interesting: better retrieval, better agents, better models, better interfaces. Those things matter, but they are not the destination. Better healthcare is.

Our work contributes most directly to SDG 3: Good Health and Well-Being, by helping make clinical decision-making more consistent, evidence-based and safe. There are broader dimensions too: better clinical knowledge infrastructure contributes to SDG 9, and making trusted medical knowledge more accessible can contribute to SDG 10, particularly where access to expertise and institutional resources varies.

We use the SDGs as a forcing function rather than a box to tick. They keep us asking whether what we are building connects to a meaningful outcome. In the end, the measure we care about is whether healthcare professionals make better-informed decisions with Elna than without it; sophisticated clinical AI is only a means to that.

Building from Barcelona, for Europe

We are particularly grateful to the Digital Catalonia Alliance (DCA) for its support. Being part of DCA-IA has connected us with other companies, institutions and people working on AI in Catalonia, and initiatives like the AI for Good Innovation Factory show how a strong local ecosystem can create opportunities that reach well beyond it.

The event brought together five very different approaches to healthcare AI with something important in common: a belief that AI can be applied to problems that actually matter. That is an ecosystem we are proud to be part of, and congratulations again to emily.AI for winning the Catalonia chapter and earning the chance to take its solution to the global AI for Good stage in Geneva.

Barcelona is where we build, and Europe is where we intend to prove this out first. It sets the highest bar anywhere for how clinical AI must be governed, which is exactly what makes it the right place to start. The experience was a useful reminder of something we believe strongly: the future of healthcare AI won't be defined by what AI can do, but by what we choose to use it for, and whether we build it responsibly enough to make a real difference.

That's the kind of AI we want to build at Elna.