I've been professionally connected to technology for 16 years and, interestingly, I've been connected to artificial intelligence in one way or another for almost that same amount of time.

When I started, AI didn't occupy anything like the media space it does today. It was a much more niche field. Back then it was used in very specific and selective environments, for example in military applications for route calculation or decision systems. In the broader industry, we were still years away from even beginning to talk about Big Data.

There was, however, a small anomaly in that map: the video game industry. And that's where I got my start as a programmer.

The early days

I began building AI for video games and real-time systems: pathfinding with algorithms such as A* and Dijkstra, state machines, decision trees, fuzzy logic, Monte Carlo methods, genetic algorithms, simulation, and emergent behaviour. They were simpler versions of ideas that were already being applied in far more demanding environments, but for me they were an incredible training ground.

On top of that, my entry into this world was partly chance and partly luck. I had a professor for whom I still have a great deal of affection and respect, who was teaching almost out of vocation after having led major international AI projects in the military sphere. He only taught for a couple of years before retiring, so I caught him by a hair. And he had a huge influence on how I came to look at this field.

Looking back, that stage gave me a very strong foundation in autonomous systems and decision-making, long before any of this became mainstream. The video game industry was quite precarious, but I learned an enormous amount.

The journey through startups

Then came different stages in companies and startups, always orbiting around AI and the major waves the sector kept going through. I moved through computer vision, machine learning, and product infrastructure, and I got to experience several of those cycles from the inside: the Big Data boom, and later its applications in areas like FinTech, PropTech, InsurTech, and mobility. Along the way I moved through engineering, technical leadership, and product roles, learning not only the technology itself, but also how to land it in very different business contexts.

Working in startups toughens you up because in small companies with limited resources, a lot is expected from you and you're exposed much more directly. That makes you grow very fast.

But the stage that marked me the most was The Cliff.

After almost a decade of working for others, I co-founded The Cliff, an AI venture studio that we built from scratch one month before the pandemic. For five years we had to do whatever it took to keep it moving forward: build product, sell, execute complex projects, build a team, refine operations, and learn how to survive while the industry was shifting beneath our feet.

I see new breakthroughs every day, but real-world application doesn't move as fast as Medium posts or AI newsletters. That's why I want to open this space.

What I'm going to share

My intention is NOT to talk about the AI trend of the week. There's already too much noise.

What I want is to share thoughtful content drawn from what we actually apply day to day in real cases: knowledge, applied AI, opinions, and the open-source projects I'm building. All of it centred on AI.

I think that many times in my career I've made the mistake of speaking too technically to non-technical audiences. Over the years I've refined that weakness, but I'm still convinced that content needs technical weight, especially now, when a LinkedIn post cooked up with a prompt can compete with one that is the result of months or years of real experience.

I want to contribute at both levels. That's why, in the posts where it makes sense, I'll divide the content into two levels:

• The business layer — the vision, the impact, the "why"
• The technical deep dive — the foundations, the architecture, or even the code behind it

Mainly aimed at business profiles, C-level, and management, but always with technical depth underneath for anyone who wants to dig further.

Liminal

And to give these ideas a more serious home, I've created a blog called Liminal.

I chose that name because it comes from the Latin līmen, meaning threshold: the point between two states. A liminal state is that moment when something is no longer what it was, but has not yet fully become what it will be. And in AI, we are constantly there.

Systems evolve faster than we have time to define them properly. Architectures shift before they stabilise. What works today is already being replaced tomorrow by another layer, another pattern, or another way of working. We are not operating on stable ground. We are operating at the edge.

Liminal is meant to be precisely that space: a place to explore systems, architectures, and ideas while they are still taking shape, before they become standard. Between the new trend of the day and the technology everyone already knows how to master, there is a far more interesting phase: the moment when you try to turn what is new into something truly applicable and valuable in real environments. That's where I want us to move together. Will you join Liminal?