Ten fill-in-the-blank prompts that take an idea from a market scan to a 30-day MVP test.
Take a prompt that worked, mark the parts that were only true that day, and keep it somewhere you will still find it in month 3.
Why one enormous request buys a thin answer, and the 5-step sequence that fixes it.
The 3 asks that replaced "think step by step", and how to read what comes back for the parts you can actually check.
Install Codex, scope its access, switch on Computer Use and save your rules in AGENTS.md, so it starts every task already set up.
One finished sample settles the format that two rounds of adjectives could not.
Role, context, task, format and constraints - what each one actually buys you, and which two you are probably leaving out.
The 5 uses that come round every week, a rhythm under an hour, and the rule that keeps it running in a bad week.
Talk to ChatGPT, Claude or Gemini the way you would brief a colleague, and give it the full picture on the first try.
Write the four-part standing instruction once, scope the rest to one project, and stop re-explaining yourself at the top of every chat.
Capture your voice once from three things you already wrote, then check any draft in twenty seconds before it goes out with your name on it.
The account you are signed in to decides most of it, and four words tell you what never goes in the box.
Why a confident answer can be false, the 4 shapes the invention takes, and a 30-second check tied to what a mistake would cost.
Paste it or attach it, what each format really gives you, and how to tell what your assistant did not read.
The Think button trades time for care, not intelligence. Four questions that tell you it is worth the wait, and what each assistant calls it.
Three parts turn a vague ask into one an assistant can follow, and four old prompting tricks you can stop typing.
The first reply is a draft, not a verdict. Four lines that fix most of what comes back, and the point where a new chat beats another correction.
One question sorts every task on your list, plus the four kinds of work that pay off in your first week and the four to keep.
What each model is built for, what the launch benchmarks actually say, and the job each one earns on a working day
Open the account, find the three parts of the screen that matter, and have one real conversation in about ten minutes.
A one-sentence definition of what an AI assistant actually is, and a five-line rule that picks yours in under a minute
What Astra does differently, and how to prompt it so it actually does the work
Grouped by what you're doing and filtered by what you already know - with why each one matters
Turn one prompt into a working app with user accounts and paid subscriptions, no code and no developers
Build a real-time AI voice agent that sounds human, knows your business and never misses a call, in six steps with ElevenAgents
