Say where the answer will live.

A clear ask still lands wrong when the model writes for the wrong surface. Name where the answer will live in the prompt.

Prompt Fluency is communicating so the model doesn't have to guess. Where the answer will live is one of those guesses you never notice making. You already know it goes on a whiteboard, in a group chat, or on a door. The model only has your ask. So it picks a default surface: a neat little essay.

Try this ask:

"Write a note about tomorrow's lineup for the little league team."

You get three paragraphs, a pep talk, and a closing line about "giving it our all." Fine if this were an email newsletter. Useless if it has to fit on the dugout whiteboard in marker while kids are already putting on cleats.

The model didn't ignore you. It wrote a note. It just invented the place that note would live.

Same break shows up everywhere.

"Draft something about the black backpack in the lobby for the apartment building."

"Make a kiln reminder for the pottery studio."

"Write the handoff for the ice rink snack bar."

Something. Reminder. Handoff. Every one of those assumes a surface. Email, flyer, text, door card, sticky note. Leave the surface out and the model picks the average one: polite, padded, and hard to use in the real spot.

Name the place right in the prompt. Plain English. One line.

"Write a note about tomorrow's lineup for the little league team. This goes on the dugout whiteboard in marker. Max 6 short lines. Names and positions only. No pep talk."

"Draft something about the black backpack in the lobby for the apartment building. This is a building-wide email from the front desk. Subject line plus 3 sentences. Include where to pick it up and that we need it claimed by Friday."

"Make a kiln reminder for the pottery studio. This is a card taped to the kiln door. Two lines max. Say the load finishes at 6pm and do not open early."

"Write the handoff for the ice rink snack bar. This is a note the morning closer leaves for the afternoon opener. Bullet list: what's low, what to thaw, and one broken machine. No full sentences if a bullet works."

Look at what changed. Same job. Same facts. But now the model knows the room the words have to survive in. A whiteboard does not get a paragraph. A door card does not get a subject line. A handoff note does not need a greeting.

No shame here. I do this too. Once you know where something goes, it stops feeling like information. It feels like the air. The model is not standing next to that whiteboard with you.

A quick check before you hit send:

Places that need a line in the prompt

• Physical surfaces: whiteboard, door card, counter sign, sticky note
• Message surfaces: group chat, email, text, voicemail script
• Shared docs: checklist, shift log, handoff note
• If a stranger walked up and asked "where does this go," and you'd point at a real spot, name that spot for the model

When the answer comes back close but wrong, ask yourself: did I name the place? Odds are the model wrote for a different room than yours. Name the room. Ask again.

You're not being picky. You're removing the need to guess. That's Applied Intelligence Mastery in practice: say where the answer will live, so the model writes for your surface instead of the average one.