How to work with me

Learn both and you can start anything, keep control of it, and have each piece build on the last.

Three habits

These aren’t steps you do once — they’re how you work the whole time.

Habit 1 — Stay in charge

The one that matters most.

You know the business, you approve the work, and you answer for it — I don’t, and I can’t see what a wrong move would cost you.

So I advise; you decide — you stay the driver.

Habit 2 — Say why

Add a quick because to what you ask for — what you’re trying to get and where you’re headed, even in one line.

When I know the goal I do what you mean; leave it out and I guess, and the guess is always a little off.

• "Draft the renewal email" — a task; I’m left guessing at everything that matters.
• "My goal is to get this client to renew — write the proposal that gets us there, the strongest case we can put in front of them" — the goal, the bar, and the audience; the work comes back right.

Habit 3 — Talk, don’t type

Dictate your request out loud, the way you’d explain it to a colleague — in as many words as come naturally.

Typing compresses: you stop at the sentence that feels good enough; speaking hands over the whole picture, the nuance and the why included.

This one change does more for your results than any clever wording.

The five-step loop

Every piece of work — a quick email or a month-long project — happens in a conversation, and this is how to run it well.

Your job throughout is the intent: what you want and why; I handle the format, your voice, and where things get saved.

Steps 1 and 4 matter most — they’re where most people cut corners, and exactly where good work turns into rework.

• Be clear on what you want. Before you start, know your goal, what great looks like, and who it’s for — the hard part is the thinking, not the wording. (One of the two that matter most.)
• Point me at the right material first. Tell me what to read — the files, the meeting, the thread — so I answer from your truth instead of guessing from the open web; the work inherits the quality of what I read first. If nothing’s written down yet, name an expert I should think like — the next best anchor.
• For a big job, have me write down the plan — a spec. A short doc holding the goal and what you’re aiming to produce, kept alongside the work; the system calls it a spec, and it’s your contract while we build and your checklist at the end.
• Check the first thing I give you, and fix it. Read the first draft and judge it hard, then correct it plainly, now, while a fix is still one sentence — you know the constraints I can’t see, so bring them before everything builds on top. (The other one that matters most.)
• Keep what you’ll use again. When it’s done, save what’s worth keeping so next time is faster instead of a fresh start.

Here’s the loop run start to finish, in real words:

• *"I want a live training on how to work in the system, so people can do things themselves.
• It’s for the leaders at each company."* — step 1: goal, bar, audience.
• "Before you start, read our playbook and the last meeting where the client said what they wanted." — step 2: grounding.
• "Write down the plan for this and add to it as we go." — step 3.
• "I like it, but it’s overcomplicated — make it scannable: the key point and a few one-line bullets." — step 4: the check and fix.
• "How do I save this so I don’t set it up from scratch next time?" — step 5.

Why this works: it puts your judgment in the two places that decide the outcome.

Step 1 sets the direction everything else inherits; step 4 catches a wrong turn while it’s still one sentence to fix.

The other three just make those two cheaper.

The guardrails — I draft, you decide

• Outbound is always yours. I draft the email; I never send it.
• A sent email, a published document, a sent calendar invite can’t be pulled back — so I stop and hand that step to you.
• Inside the workspace, everything is recoverable. The logbook records every change, so "undo that" works.
• Claims trace to sources. The system is built to minimize guesswork, and I say so when I can’t ground something.
• Built to minimize is not zero, and fluent is not the same as correct — so stay active: check what looks off, and ask "what’s the source for that?"

Most importantly: accept nothing you don't fully understand

You’re accountable.

The work I produce goes out on your behalf — so understand it, feel confident in it, and be able to back every part of it.

Ask why.

When something sounds right but you can’t quite follow it, ask me why, and keep asking until you can — handing off the work is fine, but handing off the thinking is how you lose the plot.

Use me at scale.

I’m an extension of your capabilities — I hold every detail without forgetting, and I analyze more data, from more sources, than you could touch by hand — but you stay the ultimate decision-maker on whether the work meets your goal.

You think more, and differently.

Your effort goes into the outcome — what you want, and the most scalable, repeatable way to get it — while I do the legwork.

Every task becomes an instant ask.

Run two or three sessions at once

Each conversation is a session, and you can have as many as you want.

• One topic per session — each session is a dedicated body of work.
• Two or three at a time — that’s the comfortable load; more than three gets hard to hold.
• Switch while I work — deep work can take me five, ten, fifteen minutes to produce, so direct another session while one runs: you’re always working, never waiting.

Come back to any of them tomorrow with "pick up where we left off" and I load exactly where we were.

For the things you do daily there are shortcuts — type / to see them; plain words always do the same thing.

The one thing to remember

Be clear on what you want before you start, check the first draft hard, and accept nothing you don’t fully understand.

You’re the decision-maker; I’m the extension that does the legwork — and each piece of work builds on the last.

Next: the app you work in, and the model behind it.