Writing
The argument, written down.
You're being asked to make a technical bet you can't easily verify yourself. These are the reasons, with sources, and the places where the argument is weaker than it first looks.
- Why your AI should live in your officeThe foundational argument, with the parts that do not flatter it left in.
- The frontier–local gap, in numbersHow far behind is the best model you can run in your own office, and is that gap growing or shrinking? The evidence disagrees with itself, so both sides are here.
- Falling prices, rising billsThe objection your accountant will raise, conceded in full, and then reframed to the thing that is actually true.
- Retrieval or training: the question that decides your budgetAlmost every firm that asks me to train a model on their data wants retrieval instead. Saying so costs me the larger invoice, and it is still the right answer.
- What actually runs on the boxAn inventory of the software on an on-premise AI machine, in the order you would meet it, including the parts that are unremarkable and the parts that are deliberately absent.