How to prompt AI.

A prompt is how you point the AI at what you want. It is not just the sentence you type; three things shape what you get back:

• Your message: the task you are asking for.
• Your saved instructions: preferences you set once in settings, like "keep answers short" or "I run a small landscaping business."
• What it can see: attached files, pasted screenshots, connected apps, and anything you tell it to go find.

That is why two people can type "review my budget" and get completely different results. One attached the spreadsheet, or said where to find it. The other attached nothing. Same words, completely different prompt.

Before you send a prompt, ask yourself: would a colleague reading this with no other context know what to do? If not, the AI will not either.

The takeaway: give the AI the context it needs, and treat it like a brilliant but new employee.

Here is a task someone at a nonprofit might face. The vague version: "Write a thank-you email to the sponsor." The AI has to guess which sponsor, what they did, what you want next, and how your organization sounds. You get something generic, then spend three messages fixing it.

Now the strong version: "Write a thank-you email to the company that sponsored our fundraiser dinner last night. My notes from the event are attached. Thank them for the $5,000 sponsorship, mention that we hosted about 200 guests, and invite their team to visit one of our programs this fall. Keep it under 150 words, warm but professional. Don't ask about next year's sponsorship yet."

Same task, four changes: the situation is named, the AI can see the real notes, done is defined, and one boundary is explicit. That structure is repeatable. Here is the template:

What I want: [the task and the outcome, in one or two sentences]

What you need to see: [attach the files, paste the notes, or name the source the AI should use]

What done looks like: [format, length, audience, and anything that must be included]

What to avoid: [anything off-limits, plus mistakes you have seen AI make on this before]

Before you act: [for big or risky tasks: "show me a plan first and wait for my go-ahead"]

The above is not a checklist, just the shape of a good prompt: skip any line that does not apply. Quick questions do not need prompts like this, but for bigger tasks, filling it out takes 30 seconds and should save you from correcting the AI many times later.

The takeaway: make sure to give the AI the complete story in terms of what you want.

For this issue we analyzed about 2,800 of my real prompts to Codex and Claude from the past ten months. Three techniques showed up constantly, and they matter most with agents: AI tools that open files, use apps, and keep working for minutes, even hours.

• Make it plan before it touches anything. My most repeated line, by far: "Don't make any changes right now. I just want you to plan and think with me." The AI proposes a plan, I approve, then it acts. For you: if an AI can change real things you care about, make "show me the plan first" your default opener. The tools support it directly: Codex has a /plan command and Claude's agents have a plan mode.
• Point, don't paste. I almost never retype context. I mention files on my computer, tell the agent to pull my notes through a connector (a connected app; ChatGPT calls them apps), or capture the window I am looking at. In the ChatGPT desktop app on Mac, where Codex now lives, tapping both Command keys sends the app in front of me as context (Appshots). For you: attach it, connect it, or name it in the prompt.
• Dictate your request instead of typing it. I speak most prompts nowadays. On a Mac, press Command+D to dictate inside Claude. In Codex, press Ctrl+Shift+D. If your shortcut differs, click the microphone button. For every other app, I personally use Wispr Flow. My dictated prompts tend to ramble, but that is fine, because the rambling is context: what happened, what I already tried, what I am worried about. This tends to be far faster than if I were to type my prompt.

One big tip: when you catch yourself repeating the same setup in every prompt, save it as a custom instruction, a project, or a skill so it is already loaded next time. Step 5 of the full guide explains this further.

Note: none of these are about clever wording. They are about being intentional about what the AI sees and what we want it to do.