Hi, I'm Dario
Hi everyone, I’m Dario Sansano.
I’ve always been that kid (and now adult) who can’t stop thinking about how systems work, technical ones, social ones, even political ones. I grew up building small automated things just for fun, then got lost in 3D modeling, code, and later… law. Yeah, a weird mix.
A few years ago, I ran into transformers and AI, and something clicked hard. Not in a “wow, cool tool” way, but in a this changes how power, knowledge, and value are distributed kind of way.
What bothers me today is how much of AI feels like renting intelligence instead of building it. You use it, you depend on it, but you don’t really control it. That question has been living rent-free in my head for a while.
I’m currently building NeuroBlock, but more than a product, it’s my attempt to explore a different path: one where data, models, and intelligence are things people can actually understand, shape, and own.
I’m here to learn, share ideas, and have conversations about where all this AI is going, with people who are either building or heavily consuming AI in their lives and businesses.
Happy to connect.
Replies
Hey Dario, sounds pretty cool. Just wondering - is NeuroBlock something that would run locally on a user's own hardware in their own private network? Or is it something that runs on your cloud service.
NeuroBlock
Hi Robert, great question. Right now, we’ve launched a cloud-based version where the full pipeline, data ingestion, structuring, training, and deployment, is integrated into a no-code visual interface that anyone can use. Even so, users can download the model weights and fully own the models they generate. That said, our end goal is not the cloud.
We’re actively working toward making everything runnable locally or deployable on the user’s own infrastructure, without us touching or intermediating their data or models at all. This includes research into model compression and optimization, so these systems can run on as many personal machines as possible, not just large servers.
In parallel, we’re also exploring the idea of a dedicated AI computer: a finished, self-contained machine with NeuroBlock preinstalled (there is already gret hardware in the market for this). Think of it as a personal AI workstation where you can dump and automatically structure your data, fine-tune models, use RAG, and run inference locally, fully under your control.
At a higher level, I see AI as the natural evolution of the personal computer. Right now, though, the “personal” part is missing. The industry is chasing scale and a hypothetical AGI in distant data centers, and I’m honestly skeptical that path gets us where people think it will.
What makes far more sense to me is a plurality of expert AIs, built and owned by domain experts, tailored to real problems, rather than a single generalized intelligence that everyone rents from somewhere else.
That’s the direction I'm trying to explore with NeuroBlock.