Yessica Klein
8 min read
09 November 2025
When Angelo Dalli began studying artificial intelligence at the University of Malta in the mid-1990s, he often had to justify why anyone would dedicate their life to such an abstract field. In 1996, AI was not a buzzword or a global investment magnet; it was a niche academic subject that many people dismissed as futuristic speculation. “Back then, it wasn’t obvious why you’d study AI,” he says. “We were a bunch of optimists, maybe a little bit too early – three decades early. Now that choice has been vindicated.”
His first venture into entrepreneurship was mailing software on floppy disks to customers around the world in the days before downloads existed. That instinct to see opportunity in uncharted spaces set the tone for the rest of his career. Now based between London and Malta, Angelo is a serial founder with nine startups behind him. His latest venture, Umnai, is developing hybrid intelligence: an approach that combines neural networks with symbolic reasoning to build transparent, trustworthy AI for mission-critical contexts. Alongside his technical leadership, Dalli has published more than 23 peer-reviewed papers, filed 52 patents since 2019 and remains active in the angel investor community.
You’ve founded nine companies. How has your entrepreneurial approach evolved?
All my life, I’ve been selling software. My earlier startups were in more established markets where demand was already clear. Umnai, founded six years ago, was different: when we started, AI wasn’t mainstream, and I had to explain it to almost everyone I met. Experience has taught me to be more disciplined: time-boxing my day, hiring carefully and always having an exit strategy in mind. Startups will always have moments where you think the sky is falling – persistence makes the difference.
How does Umnai’s technology differ from current AI systems?
Most modern AI, like GPT, is built on neural networks that predict the next word or outcome statistically. They’re powerful but unreliable: they hallucinate, they can’t be fully trusted. At Umnai, we’re building hybrid intelligence by combining neural networks with symbolic reasoning – what we call neurosymbolic AI. This means every decision can be traced step by step. If you’re writing a poem, maybe you don’t need that, but if you’re flying a plane, making a healthcare decision or moving millions of dollars, you need AI you can rely on. Our focus is on mission-critical systems where the cost of error is high.
What has been your biggest learning as a founder?
That startups are never a straight line. There will always be crises. Persistence is crucial, but so is discipline. I’ve also learned not to over-engineer at the beginning – keep the scope simple and build something usable quickly. Hiring is another big one: early decisions shape the culture, so adding people just to scale fast can be a mistake. And finally, financial oversight is key. Having someone constantly scrutinizing the cash flow allows innovation without losing control.
How do you see Malta as a place to build startups?
Malta has its advantages: strong incentives, competitive IP treatment and government support through agencies like Malta Enterprise, which was especially important during the pandemic. When I started out, there was almost no startup community – it was nearly impossible. Now events are full, there’s real interest and even international colleagues enjoy coming to Malta for company meetings. The ecosystem is smaller but more connected, and that sense of community is a strength.
“In Malta today, whenever there’s a startup event, the room is full. Ten years ago, that was unimaginable.”
How has remote work changed the way you build companies?
Remote work has transformed startups in ways I couldn’t have imagined when I launched my earlier ventures. Back then, it was almost inconceivable to run a mostly remote organization. Now it’s the default: our team is distributed, collaborating seamlessly across locations. What’s even more interesting is how AI has become part of that workflow – our developers use AI tools to accelerate coding, so in a sense we’re using AI to help build more AI. That has made us far more productive and allows us to tackle problems that would have been impossible just a few years ago. Combining this flexibility with our presence in both London and Malta has been a real advantage. London gives us access to global capital and markets, while Malta provides strong incentives, support and a growing ecosystem. The combination of a distributed team and a dual base in the EU and UK lets us move faster and adapt to a changing environment.
How have mentors shaped your journey?
I was fortunate to be mentored by leading figures in AI, like the late Professor Yorick Wilks, a pioneer in machine translation. Good mentors give you a sounding board, they spot blind spots. From the business side, I’ve also had guidance from other founders and investors. I think you can often judge a founding team by the quality of mentors they bring on board.
What advice would you give to early-stage founders?
First, be persistent. Don’t give up when things get hard – that’s the moment that defines success or failure. Second, build a support system: accelerators, coworking spaces and communities. Founders can seem strange from the outside, so it helps to be surrounded by like-minded people. And finally, visualize your goals and keep your eyes on the prize, even if the path changes along the way.
“A bit of knowledge of art, of perspective, of how colors change in daylight or nighttime, can actually help in fields like computer vision.”