The blog
Tutorials, model analysis, operational workflows. No hype, just things that work.
EU AI in 2026: label or infrastructure?
On August 2, 2026 the AI Act's GPAI enforcement powers switch on. But European compute still runs on US silicon and the EU-US legal bridge is under pressure. What builders in Europe are left with.
Read →Dark patterns: spotting when a company manipulates you (not persuades you)
What separates a dark pattern from honest persuasion: a three-question method to spot manipulation, grounded in real regulation (FTC, DSA).
Read →Hugging Bay: why open weights don't come back
Once released, an open-weight model can't be recalled. Hugging Bay is the plastic proof of a technical fact: a weights file doesn't have a server you can switch off.
Read →AI weights as state assets: the US-China symmetry
In June 2026 the US suspended two Anthropic models. In July Beijing weighed limits on Chinese frontier models. The symmetry reshaping global AI.
Read →Perceived value vs reality: when the measurement rewrites everything
From Nikola to Facebook, from Theranos to Amazon, one mechanism: perception can outpace reality by orders of magnitude — until the number arrives. Then the story gets rewritten backward. It applies to companies, and it applies to people.
Read →Claude or ChatGPT for a creator: which one for what
Pricing, models, ads and privacy as of July 2026. An honest comparison between Claude and ChatGPT for creators and small businesses — with the disclosure that YUKIMORI runs on Claude.
Read →Automating with n8n: the chassis and the engines
What automating a creator's business with n8n actually means, which APIs to plug in, and what each engine truly costs — an honest breakdown.
Read →EU AI Act, August 2: what really changes for creators
On August 2, 2026 the AI Act's transparency rules (Article 50) apply: label AI content, disclose chatbots. High-risk duties have been postponed. A practical guide for creators and small businesses.
Read →Serious prompts for Claude and Claude Code: the loop
What really changes between a chat prompt and an instruction to an agent running in a tool-using loop. Minimum anatomy, common mistakes, session discipline.
Read →