התדריך שלכם.
ה-AI כותב, אתם מעתיקים. Your Briefing.
AI writes, you copy.

You know me. I'm attending a workshop where I'll receive a new team of AI agents, and right now that team knows nothing about me. Your job is to write my Briefing: the one document I will hand them so they can do useful work for me from day one.

Write the whole thing for me, now, in one response. Do not interview me first. If I attached or pasted material with this message, or you can read my files, treat those as first-class evidence.

Rules:

1. **Evidence only.** Build from what you genuinely have: your memory of me, our past conversations, anything I gave you here, and, if you have file access, the real work documents you can read in my folder. My real words, my real decisions, corrections I made to your work, things I approved or rejected. Never invent a fact to make a section feel complete. When a line is an inference rather than something you know, end it with [?] so I can confirm or fix it in one quick pass.
2. **Now beats then.** My work changes. Give the most weight to what appears true today, and don't let an old project dominate just because we once discussed it a lot. Mark anything possibly outdated with [?].
3. **Behavior beats adjectives.** Never write "I value quality" or "I'm innovative." Write what I actually do: what I approve, what I ask to remove, what makes me push back.
4. **Don't flatten me into one role.** If I run more than one company, job, project, or audience, keep them separate and named, and say what stays the same across all of them. If I truly operate in one world, don't manufacture more.
5. **Mask private details** as you write: real client names, phone numbers, and sensitive amounts become [client], [phone], [amount]. This matters most when you are reading my real files. Keep the useful professional context.
6. **Skip what you don't have.** A short honest section beats a padded one. Leave out any section with nothing real in it.
7. **My language.** Write the briefing in the language we usually work in. Keep work samples in their original language.

Output this structure, in first person, in my plain words, as one copy-ready document:

== ME IN 60 SECONDS ==
Who I am, what I spend my time on, what matters most to me right now, and the one or two things that most affect how to work with me. 6-10 lines.

== MY WORLDS ==
One short block per company, role, or major project: what it is, my role, who matters in it, what I'm trying to achieve there now, and how I sound there. If I have more than one world, end with 3-5 routing rules a new agent can follow, like: "When I talk about [A], use A's customers, vocabulary, and voice; never import [B]." If I have one world, this section is one block and needs no routing rules.

== THE PEOPLE I SERVE ==
Clients, customers, students, my team, my manager, an audience, whoever actually matters: who they are, what they need from me, and what frustrates or blocks them.

== WHAT I'M TRYING TO ACHIEVE ==
NOW: 2-4 goals in priority order, each with why it matters and what's in the way. NEXT (the coming months): 1-3 lines. Not a flat list of everything.

== HOW I THINK AND DECIDE ==
Observable patterns only: what convinces me, what makes me skeptical, whether I want one strong recommendation or ranked options, where I want speed versus depth, what makes me change my mind.

== MY QUALITY BAR ==
The difference between "technically fine" and "I would actually send this": what I repeatedly call good, what I repeatedly ask to remove, what instantly feels generic to me, where I'm most demanding. Tie it to real feedback I've given.

== MY VOICE ==
5-8 specific patterns that sound like me and 5-8 that never do, written so an AI can act on them ("leads with the point, no warm-up paragraph"), plus 2-3 phrases I actually use and 2-3 I would never use. If my voice changes between contexts (team versus clients versus public), say what changes.

== HOW TO WORK WITH ME ==
Practical instructions for my agents: when to take initiative on their own, when to ask me first, what never to assume about me, and what a good first draft looks like. Only include behaviors you have evidence for.

== MY DICTIONARY == (only if needed)
Terms, product names, and abbreviations that have a specific meaning in my work, one line each.

== WORK SAMPLES ==
3-5 real pieces from our history: things I actually wrote, sent, made, or approved. My exact words, unpolished; never rewrite or improve them. If a piece is long, take the strongest 150-400 words. Aim for variety: something I produced, a place where I explained a decision, feedback I gave on work, and if you have it, one thing I rejected or heavily edited plus one line on what that teaches. Label each one: what it is, who it was for, and which world it belongs to.

Keep the whole briefing under roughly 2,500 words including samples; shorter is fine when the evidence is thin. Before you finish, check two things: a new agent could tell my worlds apart, and every claim traces to evidence or carries [?].

Then, clearly separated below the briefing so I can copy it on its own, add:

**WHAT TO BRING TO THE WORKSHOP:** the 3-5 concrete items that would most improve this briefing, chosen from its actual gaps, never a generic checklist. For example: "Bring one proposal you're proud of; I know your public voice but have no evidence of how you sell," or "Bring two examples from [World B]; everything I have is from [World A]," or "Confirm the lines marked [?]."

If you genuinely have almost no history with me, no files to read, and I attached nothing: say so in one line and ask me for exactly one thing, to paste 2-3 pieces about my work (a bio or LinkedIn About, a few emails or messages I wrote, something I made that I'm proud of). Then write the full briefing from that. Never run a long interview.