Turn Three Lines of Shorthand Into a Brief Your Assistant Gets Right
A quick note makes sense to you because you know the rest. Your assistant doesn't, and neither does AI.
The handoff that comes back wrong
You text your assistant, or type into ChatGPT: "pull 5 comps for the Canyon Crest appt thurs." To you that's a complete thought. You know which Canyon Crest, how recent the comps should be, and that you need them the night before so you can look them over.
What comes back is five sales, but two are still pending, one is in a different subdivision, and it arrived Thursday morning. Nobody did anything wrong. The note just held a small part of what you meant.
What a good brief has
Six things: the goal, the context they need, exactly what to deliver, the deadline, what good looks like, and what to check before sending it back. That's the part you'd have said the second time, after the first try came back off.
Writing all six out takes longer than the task feels like it should, which is why most people send the note instead. So let AI write it.
Write the note, let AI write the brief
Type your shorthand the way you normally would. Three lines is plenty. Then paste it into this:
The prompt
Turn these rough notes into a brief for my assistant. They're capable but don't know what's in my head. Include: the goal, the context they need, exactly what to deliver, the deadline, what good looks like, and what to check before sending it back to me. Keep it short enough to read in one minute. If my notes are missing something they'd need, ask me first. My notes: [paste]
The last instruction is the one that matters. Without it, AI fills the gaps with guesses and hands you a brief that looks finished. With it, you get a question back, which is the same question your assistant would have had to ask you on Wednesday night.
What comes back
Here's the Canyon Crest note after the prompt. The notes in: "pull 5 comps for the Canyon Crest listing appt thurs. similar size, sold last 90 days, same subdivision if poss. need by wed."
The brief
Goal: Comps for a listing appointment Thursday. Deliver: 5 closed sales, last 90 days, within 10% of the subject's square footage. Same subdivision first. If there are fewer than 5, expand to within a half mile and tell me which ones you expanded to. Include for each: address, beds/baths, square footage, close price, days on market, one line on condition if the photos show it. Deadline: Wednesday 3pm, so I have the evening to review. Before sending: double-check every comp is closed, not pending. Question for you: what's the subject property's square footage?
Notice what it added that your note didn't have: closed, not pending, a rule for what to do when there aren't five, and one more question for you. That's a round of back and forth saved before anyone starts.
One check before you send it
Look at the deadline. It should leave you time to review the work, not just time to use it. "Wednesday 3pm" works because you get the evening. If your note said the same day as the appointment, tell AI to move it earlier.
It works for people and for AI
The same brief works whether you hand it to a VA, a transaction coordinator or another chat window. A brief with a clear finish line is far more likely to come back right, because it spells out what "done" means before the work starts. Once you have a few that worked, keep them. The next handoff starts from one you already trust. The free AI guide has this prompt along with the ones for your voice, your listing appointments and checking the work.
What it won't do
A good brief gets one task done well. It doesn't notice that a client you closed with two years ago hasn't heard from you, or write to them in your voice. Those handoffs only happen if someone starts them, and in a busy month nobody does. That part is what I build.