Hi, I'm Barbara. Want to know why this blog exists?

There's an hour, late at night, when the rest of the world stops asking me for things. That's when my second job begins — because by day I have one, like everyone else: I work as a TV crew assistant. At night I open my laptop and try to teach a machine to imagine what's in my head. I live a kind of half-life, and for a long time I lived it as a lack. Then I understood it was the most mine part of me.

I didn't come to artificial intelligence as a technician. I came to it as a creative — a head full of images and almost no time to make them real. For me, AI wasn't the magic shortcut those thirty-second videos promise. It was something to be learned by hand: getting it wrong, redoing it, until the machine starts to answer you back. And like with any serious tool, the difference isn't the gear. It's the person using it.

I've spent the last three years treating it like a craft, not a toy. I analysed more than ten thousand prompts and tested over three hundred tools — not to collect them, but to understand why sometimes you get an image that looks real and sometimes a caricature with six fingers. The answer was almost never in the model: it was in what I had asked for, and above all in how I had asked.

Why I opened this site

This blog comes from a very specific irritation: the idea that something becomes automatically simple just because it was made with AI. Too many people sell it as a magic wand, and too few show you the real work — the bad drafts, the prompts rewritten ten times, the photos thrown away. I wanted an honest place to write down what actually works, with the names of the tools and the steps, without pretending it's all easy.

It isn't a diary, and it isn't a course: a course has a syllabus, a beginning and an end, and when I make one I'll tell you plainly. This is more like taking you onto the set: showing you what lies behind the glossy image of the finished thing — the failed attempts, the things that go wrong, the weeks of preparation it takes to make a few seconds work. If you come back, you'll find the method as it forms: what I understood this week, what I got wrong, which tool I stopped using and why. I write in the first person because it's the only way I know how to be useful: telling you what I've touched with my own hands, not what I read somewhere.

The rule I've given myself is simple. If I haven't tried it, I don't write it. If I can't reproduce a result, I won't sell it to you as method.

What I actually do

A big part of my work is teaching AI to people who create content, and to small businesses that can't afford a marketing department. I build reels in which my character performs inside carefully made scenes, and at the end of the video I let you "steal" the prompt: the exact recipe to remake it. It's my way of showing how it's done, not just what comes out. The thing that changed my work wasn't finding the most powerful tool, but realising that you can build new, custom tools by wiring together the AIs that already exist: that's what speeds up the creative flow, and that's where a system comes from. And a system, disciplined by a method, scales better than a talent in the long run.

Behind a fifteen-second reel there are often hours of invisible work: the voice built to measure, the images redone until the light is believable, the consistency from one shot to the next. I use tools like Seedance to animate the scenes, Nano Banana for the images and ElevenLabs for the voice, but I name them the way you name brushes: they matter, yet the hand comes first. The hard part isn't pressing "generate." It's knowing what to ask for, and recognising when what came out isn't good enough.

The real work, and the proofs of strength

Part of what I do comes from real commissions. I created the communication for Orange1's anniversary, and I work with AreaVideo, a company active in video and communication for over thirty years: there the constraint is the same old craft — understanding what the client needs and delivering it, with AI as an ally and not an alibi.

Another part, though, has no client at all: I invent it. I can invent an entire campaign — and tomorrow, probably, something else. AMALUR is my favourite example: a cosmetics brand that doesn't exist. A campaign built to prove — to me before anyone else — that with these tools you can hold up an entire brand identity from scratch.

What you'll find, coming back

You'll find three kinds of things, braided together. Method: short, concrete pieces about something I learned using AI in my work, with the example right in front of you. News: when something happens that changes the game for people who make content, I'll tell you about it and what I think, with the sources in the open. And the more personal pieces, where I try to line up the meaning of all this — why I spend my nights teaching a machine to bring my ideas to life faster, without losing my own creative identity along the way.

I won't promise to make you an expert in an afternoon. I promise you something more useful: to show you the work as it is, mistakes included, so you can truly steal it. Because the truth is that AI doesn't replace us. It only asks us to get better at saying what we want.

The rest I'll write as I go. Come back whenever you like: what you find here is the work in progress, not the finished, polished result. Because the only way to use every byte of this technology is to keep progressing. With no finish line.