-
Why 'Alignment' Is the Wrong Word for What We're Actually Worried About
AI alignment sounds like engineering. It isn't — it's political. The word hides this, and the safety field has been paying for that mistake ever since.
-
I Tracked Every AI Query for a Month — The Results Changed My Setup
I logged every AI query for 30 days using Open WebUI logs and the llm CLI. The data showed 80% of queries run fine on a local model — here's the routing setup and cost savings that followed.
-
Local LLMs with Ollama: Thermal budget of a MacBook pretending to be a server
A field-tested take on the thermal budget of a MacBook pretending to be a server with Local LLMs with Ollama: what it rewards, where it breaks, and how to keep the workflow honest.
-
Invisible Details That Separate Premium Products From Average
Discover the hidden details that make premium products feel special. From material choices to interaction design, learn what separates exceptional.
-
France Had the Best AI Research in the World in 1990. Then This Happened.
France had world-class AI research in the 1980s. How it lost that edge — and whether Mistral can reverse it — is a lesson every country should study.
-
The Spreadsheet Killer — Local AI Data Analysis on Your Mac
Natural-language data analysis with local LLMs on Apple Silicon. Why LLMs should write pandas code instead of doing math, a local code-interpreter setup with Ollama and Jupyter, and prompt patterns for analyzing bank CSV exports privately.
-
Agentic coding: Guardrails before autonomy, always
A field-tested take on guardrails before autonomy, always with Agentic coding: what it rewards, where it breaks, and how to keep the workflow honest.
-
How Apple Tests Products in Extreme Scenarios Users Will Never See
Explore Apple's extreme product testing methods including temperature chambers, robot armies, and simulated disasters. Learn why testing for impossible.
-
Why Neuroscience Got AI Wrong for Half a Century
The brain-as-computer metaphor dominated AI research for fifty years and actively set the field back. Here's exactly how, and what replaced it.
-
AI Voice Cloning on Mac — Useful, Creepy, and How to Do It Right
How to clone your own voice locally on a Mac. macOS Personal Voice setup walkthrough, open-source TTS models on Apple Silicon, honest quality expectations, and the consent-based ethics rules for AI voice cloning.
-
MCP servers: Least-privilege auth for MCP in a real enterprise
A field-tested take on least-privilege auth for MCP in a real enterprise with MCP servers: what it rewards, where it breaks, and how to keep the workflow honest.
