Nicolo' Brandizzi – Blog 2026-07-20T00:00:00+00:00 https://nicolobrandizzi.com/feed-blog.xml Writing for Machines https://nicolobrandizzi.com/blog/writing-for-machines/ 2026-07-20T00:00:00+00:00 Nicolo' Brandizzi If you write SKILL.md files for AI agents, you have picked up a house style without being taught it: harder, terser, more commanding than the way we write for humans. I measured where it came from ... Poolside's Model Factory: Honest Engineering, and the Data It Can't Show You https://nicolobrandizzi.com/blog/poolside-model-factory/ 2026-05-27T00:00:00+00:00 Nicolo' Brandizzi Poolside shipped a strong open model in five weeks and actually told us how. A close read of the XS.2 / Laguna report: what other labs should copy, a few RL ideas it sparked, and why regulation, no... KlarText: Engineering Agents That Know When to Stop https://nicolobrandizzi.com/blog/klartext-leichte-sprache/ 2026-05-26T00:00:00+00:00 Nicolo' Brandizzi An internal project I worked on at Fraunhofer IAIS: a four-agent system that rewrites German bureaucratese into Leichte Sprache. The translation was the easy part. The engineering was keeping a loo... I Got 99 Observations and Your Bike Ain't Safe https://nicolobrandizzi.com/blog/bike-steal/ 2026-03-04T00:00:00+00:00 Nicolo' Brandizzi A statistical attack on shared combination locks. I collected 99 observations, built probability models, and cracked the code — with interactive charts. The SOTA Trap: Why We Are Working for Big Tech for Free? https://nicolobrandizzi.com/blog/sota-trap/ 2025-12-09T00:00:00+00:00 Nicolo' Brandizzi Are we doing science, or are we just unpaid QA testers? How vague 'Teaser Papers' and the 'Open-Washing' agenda are paralyzing academic research. Teaching Machines to Think: Reinforcement Learning and Reasoning in LLMs https://nicolobrandizzi.com/blog/rl-reasoning-llm/ 2025-10-15T00:00:00+00:00 Nicolo' Brandizzi A deep dive into how reinforcement learning teaches large language models to reason: from lasagna analogies to latent spaces. The Digital Poisoners: How Russian Propaganda Networks Are Corrupting AI Training Data https://nicolobrandizzi.com/blog/pravda-grooming/ 2025-07-23T00:00:00+00:00 Nicolo' Brandizzi 75GB of propaganda, 306 fake news domains, 1 goal: corrupting AI training data. How Russian networks systematically poison the web to manipulate language models. AI Gigafactories: Europe’s €20B Race and the Site Selection Challenge https://nicolobrandizzi.com/blog/ai-gigafactories-tool/ 2025-07-02T00:00:00+00:00 Nicolo' Brandizzi The EU faces a massive challenge: selecting sites for AI gigafactories. I built TerraSelect to help filter the noise and make early decisions more data-driven. Introducing PolAI Graph: Visualizing the EU’s AI Landscape https://nicolobrandizzi.com/blog/polai-graph/ 2025-01-09T00:00:00+00:00 Nicolo' Brandizzi PolAI Graph: an interactive visualization mapping the EU AI ecosystem — policies, projects, funds, and organizations in one navigable graph. AI Factories: Europe's €1.5B EuroHPC Supercomputer Plan https://nicolobrandizzi.com/blog/ai-factories/ 2024-12-17T00:00:00+00:00 Nicolo' Brandizzi Seven EuroHPC supercomputers, three funding tracks, up to 50% EU cost coverage — how Europe's AI Factories initiative actually works, and where the coordination risk sits. The Big Questions on Europe’s AI Data Policy https://nicolobrandizzi.com/blog/data-strategy-questions/ 2024-10-28T00:00:00+00:00 Nicolo' Brandizzi Training AI models in the EU? Understand the complex rules around PII and copyrighted material, and the ambiguities that remain across member states. The EU’s Data Strategy: Unlocking Europe’s AI Future https://nicolobrandizzi.com/blog/data-strategy/ 2024-10-23T00:00:00+00:00 Nicolo' Brandizzi How the Data Governance Act and Data Act shape Europe's AI data ecosystem — from open data to common data spaces and B2B sharing rules. Europe’s AI Research Agenda: Coordinated Plan and Horizon Projects Explained https://nicolobrandizzi.com/blog/eu-and-research/ 2024-10-13T00:00:00+00:00 Nicolo' Brandizzi Explore the EU's Coordinated Plan on AI, its evolution, and key initiatives like AI Factories and GenAI4EU for a human-centric, sustainable AI future. Horizon AI Fragmentation: Challenges and Budget Efficiency https://nicolobrandizzi.com/blog/horizon-analysis/ 2024-09-23T00:00:00+00:00 Nicolo' Brandizzi Is EU AI research fragmented? This analysis explores Horizon Europe project data, revealing significant overlaps and opportunities for substantial budget savings. Exploring Global AI Governance: Lessons from China for the EU https://nicolobrandizzi.com/blog/china-eu-ai/ 2024-09-18T00:00:00+00:00 Nicolo' Brandizzi Explore contrasting approaches to AI regulation. Can the EU adapt China's transparency measures without compromising free expression? Addressing Loneliness with AI : The Short-Term Solution with Long-Term Consequences https://nicolobrandizzi.com/blog/ai-risks/ 2024-09-08T00:00:00+00:00 Nicolo' Brandizzi AI companions offer comfort—but at what cost? How short-term emotional relief might deepen long-term loneliness and expose us to manipulation.