Local AI
Open-weight and open-source models and the tools that run them, tested on our own machine: how fast they run, how much memory they need, what they send home and what they cost to run against paying for a cloud service.
OpinionLocal AI privacy: how to prove your data never leaves
Most privacy is a promise you have to trust. Local AI privacy is different: here is how to prove, with…
OpinionOpen weights vs open source: what the license actually lets you do
A model you can download is not a model you are free to use. Open weights vs open source: two…
OpinionWhy we run our LLMs locally instead of calling an API
Why run an LLM locally instead of behind an API? Nothing leaves the building, the cost is fixed, and the…
OpinionOwn your AI stack instead of renting it
Own vs rent: forget the AI bill for a moment. Renting your stack also rents your uptime, your data access,…
OpinionRun Local AI Without a GPU: Start on Your Laptop
Run local AI without a GPU: the hardware wall is mostly imaginary. A normal laptop runs useful models today, and…
OpinionInvisible AI: the most useful AI is the kind you don’t notice
The AI worth having is rarely the one you open a tab to talk to. It is the quiet kind:…
OpinionLocal AI is good enough now: what we run daily
The question is no longer whether it runs. Local AI as a daily driver is good enough for most work,…
Case StudiesSelf-Hosted AI: A Studio Case Study in Leaving Subscriptions
A self-hosted AI case study: not which tool replaced which, but why we left our subscriptions, the order we migrated…
Behind-the-scenesCoding With a Local AI Agent: A Week With No Cloud
A week of coding with a local AI agent: OpenCode pointed at Ollama, no cloud and no fallback. What worked,…
Builds & ToolsThe Cheapest Mac That Runs Serious Local AI
We are asked which is the cheapest Mac for local AI more than anything else. The honest answer is not…