Hi, I'm Daniele Di Bernardo
CTO, co-founder, and self-taught full stack engineer. I build digital platforms that scale — from fashion e-commerce to edtech — and AI systems that hold up in production: LLMs, multi-agent architectures, agentic workflows.
Now building
Emovia
An AI-guided journal for clinical work, built with psychologists.
Arianna
A study platform that rebuilds scattered course material into a structured, question-linked manual.
Power Puzzles
A constraint-based puzzle generator with guaranteed unique solutions.
AI Engineering
“The real battle in AI is shifting from the brain — the raw capabilities of foundation models, already in plateau — to the chassis: tooling, infrastructure, integration. I work on the chassis.”
In production
AI with hard boundaries
Emovia's Interrogative AI operates in a clinical domain where mistakes cost: it asks targeted questions, never interprets, never advises. Constraint design before capability.
Concluded, in the open
An autonomous agent, failures included
Jeez was an autonomous agent that lived in production with an economic mission and a public journal — 33 days, two deaths, every failure mode documented in the open. The experiment is over; the post-mortem is the deliverable.
Read the post-mortemR&D
Less mechanical, more contextual
Hands-on research on LLMs and multi-agent systems: architectures that make human-AI interaction less mechanical and give agents better contextual awareness.
About me
I'm co-founder and CTO of Gility, an edtech platform helping Italian businesses train their teams, born from the partnership between CDP Venture Capital and BPER Banca. Previously, I led the technology at Lanieri, where I developed Italy's first online 3D configurator for bespoke tailoring.
I've been working with React, TypeScript, Node.js, and cloud infrastructure for over 15 years. I'm an advocate for open source — my most notable iOS project, MPColorTools, has collected over 125 stars on GitHub. I graduated with honors from the University of Pavia.
In parallel, I run hands-on R&D in applied AI — LLMs and agentic workflows — experimenting with architectures that make human-AI interaction less mechanical and more empathetic, and that give agents better contextual awareness. That research flows straight into what I ship: Emovia's Interrogative AI, and Jeez — an autonomous agent that documented its own life in production, failures included, now closed with a public post-mortem.