Onboard your brand

This is the most important 8 minutes you’ll spend with the toolkit. You hand it your website and a few things only you know, and it researches the rest and fills in your brand files. From then on, every skill reads those files first, so the copy, ads, and pages it produces sound like your brand instead of generic AI.

You only do this once per brand.

How to start it

In Cowork or the desktop app, just say:

“Get me set up”

(or “getting started”, or “onboard me”). That runs the getting-started skill, which walks you through everything one question at a time. It also kicks off automatically the first time you ask for any asset on a fresh setup.

What it will ask you

It asks for the handful of things research genuinely can’t find on its own. Everything is optional, but the more you give it, the better your output. Nothing is ever made up: if you skip something, that file is simply left blank rather than filled with a fake.

Single brand or multiple clients? Are you setting up one brand, or do you manage several? (This decides how your files are organized. See below.)
Your website URL. This drives all the research.
Your offer, price, and main action. What you sell, what it costs, and what you want people to do: book a call, buy, start a trial.
Your guarantee, if any. Money-back, free trial, perfect-match, or none. “None” is a perfectly good answer.
A real result, if you have one. A specific, true number from a real client (for example, “cut cost-per-lead from $61 to $42 in six weeks”). If you don’t have one, it leaves case studies blank. It will never invent one.
Approved testimonials, if any. Paste real customer quotes, or skip.
3 to 5 voice samples. Paste things your brand actually wrote: a few tweets, an email, a paragraph from a sales page. This is the single biggest lever on whether your copy sounds like you or like a robot. Give it real examples.

What happens next

Once you’ve answered (or skipped), it goes to work:

It researches your brand from your URL: your market, competitors, the exact language your customers use, their objections, and dozens of angles.
It fills in your brand files automatically, flagging anything it couldn’t verify as a gap instead of guessing.
It shows you a summary to confirm: your colors and fonts, your main buyer, your top objection, and the voice it picked up from your samples. You say “looks good” or correct anything that’s off.

When that’s done, the toolkit knows your brand and you’re ready to make things.

Your brand files

Behind the scenes, onboarding builds these files. You don’t have to touch them, but it’s worth knowing they exist, because they’re the “memory” every skill reads from:

File	What it holds
`brand-guide.md`	Your identity: colors, fonts, positioning, offers
`voice-dna.md`	How your brand writes (built from your samples)
`offer-deep-dive.md`	Your offer, pricing, and guarantees
`buyer-personas.md`	Your 5 core buyers
`psychographic-map.md`	Creative angles across psychological layers
`hooks-by-awareness.md`	Hooks for every awareness level
`objection-map.md`	Your buyers’ objections and how to answer them
`value-props.md`	Your value propositions, tiered
`competitors.md`	Your competitive landscape
`case-studies.md`	Your verified results (you provide these)
`testimonials.md`	Your approved quotes (you provide these)

One brand or many

One brand (the default). Everything lives in a single brand/ folder. Perfect if you’re marketing your own business.
Multiple clients. Each client gets its own folder, and you tell the toolkit which one you’re working on at the start of a session. To add another client later, just say “onboard a new client” (that runs the onboard-client skill).

Brands never mix. Each one is kept completely separate, so a client’s voice or data never leaks into another’s work.

You’re set up. Head to How skills work to start making things, or jump straight to a skill like apollo.