Inside the Machine — Episode 01
It's not the future. It's already here, and you're probably already using it.
There's a precise moment when you stop wondering if artificial intelligence is a fad or a revolution. It's the moment you realize you're already using it — and that it has already changed something in your workflow, even if you didn't immediately realize it.
That's exactly how it was for me. No epiphany, no blind leap of faith. Just a problem to solve and a new tool at hand.
The first contact
I was building the website. Factory268 was still more of an idea than a reality, and there were a thousand decisions to make — structure, content, page logic. I wasn't looking for a creative assistant. I was looking for someone to help me stay on track.
I opened ChatGPT almost by instinct. The first questions weren't technical at all: how do I organize this section, what tone should I use to describe a digital product, how does someone who sells Revit families and not shoes structure a product page. Questions I would have asked a colleague if I had one with that specific combination of skills.
And it worked. Not perfectly, not always — but enough to make me come back the next day with a new question.
The disorientation
There's an initial phase where you're not sure what you have in your hands.
The tool responds, produces text, seems to understand. But you don't yet know how far you can push it, what you can ask it, where usefulness ends and noise begins. It's like walking into a huge store without knowing what you're looking for — technically you have access to everything, but you risk leaving empty-handed anyway.
I believe most people approaching AI for the first time find themselves in exactly that store. Curious but disoriented. With a vague feeling that there's something useful in there, but without a map.
This column stems from that — from that disorientation, and the journey to overcome it. I don't have a definitive stance on AI, I don't have all the answers yet. I have an ongoing journey, and it seems more honest to recount it as it happens than to wait until everything is clear.
Finch's Machine
If you've seen Person of Interest you already know what I'm talking about. If you haven't seen it, it's worth catching up — but for now, you just need to know this: Harold Finch builds a Machine capable of processing enormous amounts of information, recognizing patterns invisible to the human eye, anticipating events. But the Machine doesn't act. It observes, processes, and ultimately delivers a number — an identification code, a name, a lead. Then it's up to humans to decide what to do with it.
That's the distinction I find most useful when thinking about AI in a professional context.
The Machine doesn't transform John Reese into someone different. It doesn't give him superpowers. It makes him more effective by providing him with information he wouldn't be able to gather himself in time. But judgment, decision, responsibility — these remain human. Always.
Applied to everyday work, this changes the initial question. Not "will AI replace me?" but "what does the Machine bring me that I can't see in time on my own?" It's a much more productive question. And the answer, you'll find, depends almost entirely on how you query it.
Three tools, three functions
My journey with AI has not been linear, and it hasn't had a single protagonist.
ChatGPT was the first contact — and remains the tool I use to quickly retrieve information, explore rough ideas, ask open-ended questions without a precise structure. It's a great starting point when you're not yet sure what you're looking for.
Gemini came in at a specific moment: I needed to bring Revit families to life visually, beyond technical views. I exported isometric views from the Revit project file, and Gemini transformed them into realistic renderings — environments where those pieces of furniture truly existed, not just floating on a neutral background. For those who work with visual products like Revit families, this type of tool has concrete and immediate value: it doesn't replace a professional render, but it reduces the time and technical barrier between an idea and its representation.
Claude arrived later, and it changed the way I work in a more structural way. Not because it's "better" in an absolute sense, but because the type of collaboration it allows is different. With Claude, I don't retrieve information — I reason. With Claude, I structure articles, build the logic of ITFs, design social campaigns, define the editorial direction of Factory268. It's the tool I'm learning to use to transform a vague idea into something publishable.
The difference, if I had to summarize it: ChatGPT helps me find, Gemini helps me visualize, Claude helps me "think." Three different functions, three different moments in the workflow. None of the three does everything — and perhaps that's for the best.
There is, however, a risk worth mentioning immediately, because it is real and underestimated. When you delegate too much, you progressively lose the ability to verify. The output arrives, it seems consistent, and the temptation is to accept it without questioning it. That's exactly where something breaks — not in the tool, but in the method. AI doesn't know when it's making a mistake. It has no doubts, no hesitations, no perception of its own limitations. That perception must remain yours. In Person of Interest this theme is central — and Samaritan, the rival Machine, is the dystopian answer to the question "what happens when the system stops being controlled?" The answer is not reassuring. In everyday work, the consequences are less dramatic, but the principle is the same: a result produced by AI must always be read, verified, questioned. It is precisely this critical capacity — the willingness not to passively accept — that distinguishes those who use the tool from those who become dependent on it.
A tool that should not be seen as a blind automaton, but a system that amplifies the reasoning of those who already know what they are looking for.
An open question
I don't have a conclusion to offer. This is the first episode of a series that will grow with my experience — and will probably change direction as that experience solidifies.
What I know, for now, is that the right approach is neither uncritical enthusiasm nor defensive skepticism. It's something more akin to patient curiosity — the same curiosity one brings to a new tool, when one understands that it's worth learning to use it well before judging it.
Finch's Machine was neither good nor bad. It was powerful. And the difference was made by who used it, and how.
See you in the next episode.