The biggest problem with AI right now is direction. Give a strong model a vague ask and you get confident work aimed at the wrong thing. The fix is to give the strongest model you have three things: what you actually want, the situation it’s acting in, and good tools. A better model with a vague ask gains you very little.
Intent, then context
Intent engineering is capturing what you ultimately want and carrying it into every task. It is still prompting. It covers the what-you-want layer of prompting, built into the system so you don’t have to retype it.
Context is the information the model needs for the task in front of it: the relevant files, prior decisions, who’s involved. Most AI tooling works hard on context. LifeOS works on intent first, because context assembled for the wrong goal is wasted.
Where each kind lives
-
Your durable intent lives in TELOS: mission, goals, problems, strategies.
-
The intent for one task lives in its ISA, written as checkable criteria before work starts.
-
The situation lives in memory: who you are, what you own, what you’ve already decided.
In practice
You type “make me a landing page for the course.” The words are short, but the system already knows which goal the course serves, who it’s for, how you write and what you deploy on. It writes those down as the definition of done before building. The prompt stayed short and the result is aimed at what you wanted.