How to Onboard a New Brand Into an AI Marketing Platform
You onboard a new brand into Orbitable by building one shared world model, not a separate briefing for every agent or every campaign. That world holds the ICP, the brand voice, the product facts, the competitor set and any uploaded knowledge base, and all 50 specialist agents across the 10 squads read from it, so the second piece of work starts with context already in place instead of a blank page.
What starting from scratch actually costs
Most AI marketing tools treat every task as an isolated prompt. You brief the tone in one tool, paste the ICP into another, re-explain the competitor landscape in a third, and every agent or workflow forgets what the last one knew. Nothing compounds. A content brief and a cold email sequence for the same client end up sounding like they came from two different companies, because in practice they did, two different context windows with no shared memory between them.
A world model removes the re-briefing step. You do it once per brand, and every agent, from the content-writer to the sales-outreach agent to the cro-builder, works from the same ICP, the same voice, the same competitor list. Consistency stops being a review problem and becomes a structural one you solved at setup.
The inputs that make up a world
A world is built from five inputs, and they are the actual object every agent queries before it produces anything:
- ICP: who the brand sells to, by segment, so agents target messaging rather than generalise it
- Brand voice: the tone, vocabulary and constraints that keep every agent's output recognisably the brand's, not generic AI copy
- Products or offers: what is actually being sold, so agents don't invent claims or pricing
- Competitors: the named set agents reference when writing comparisons, positioning, or objection handling
- Knowledge base: uploaded documents, case studies, transcripts, past campaigns, that agents can cite instead of guessing
None of these five are optional if you want agent output to be usable without heavy editing. Skip the competitor set and you get generic positioning. Skip the brand voice and you get content that reads like every other AI-generated post.
The setup sequence
The order matters less than getting all five inputs in before you expect finished output, but a practical sequence looks like this:
- Create a world for the brand. Each brand or client gets its own world, so context never bleeds between accounts.
- Load the ICP and brand voice first. These two inputs shape almost everything else an agent produces, so they earn priority.
- Add product and offer facts. This is what stops agents inventing pricing or capabilities.
- Add the competitor set. This is what lets agents write comparisons and handle objections without you briefing every single piece.
- Upload the knowledge base. Past case studies, decks, transcripts, anything that gives agents real evidence to draw on instead of generic claims.
- Hand off to the Dispatcher, which orchestrates the 50 specialist agents, or run the GTM plan, a 6-phase, 26-agent-step sequence that takes a world from research through to execution.
Once the world has enough in it, Autopilot (paid plans only) proposes a weekly plan: one focus, 3 to 5 missions each with a why-now, a carry-over list and an explicit not-doing list. It proposes only, it doesn't act on its own. Approval happens in the Dock.
Why the world compounds instead of resetting
The mechanical reason this matters: once the world exists, every subsequent piece of work is cheaper, not just faster. A content-writer draft for a campaign reuses the same ICP and voice a sales-outreach sequence used last week. A cro-builder page reuses the same competitor positioning a comparison blog post already worked out. The 22 playbooks and 9 schedule templates in the platform assume a populated world; they're pre-built sequences of agent steps, and they only produce on-brand, on-target output because the world underneath them is already filled in.
This is the actual difference between onboarding a brand into Orbitable and onboarding a brand into a general-purpose AI tool. In the first case, setup is an investment that every future agent draws down against. In the second, every task starts the investment over.
Onboarding one brand versus onboarding a client roster
If you run an agency and onboard multiple client brands, each client gets its own world, and how many worlds you can hold depends on plan.
Enterprise is custom and is the only uncapped plan. Annual billing saves 20% on any tier.
For an agency, the Dock matters as much as the world count. It's the client-facing review surface: a client sees what's waiting, marks it up with pins directly on the draft, sends it back, and approves finished work into a Library. Requests carry SLA targets, so a client raising a new brief isn't an email you have to track manually. A client seat is Dock-only, and internal team notes are never visible to the client, so your working notes and their review experience stay separate.
What to load first if you're short on time
If you can't populate all five inputs on day one, load the ICP and brand voice before anything else. Those two inputs govern tone and targeting across every agent, and most first deliverables, a content piece, an outreach sequence, a landing page draft, will read as on-brand even before the competitor set or knowledge base is fully built out. Add competitors and documents as you go; nothing forces you to redo the ICP or voice setup when you enrich the world later.
FAQ
How long does it take to onboard a new brand into Orbitable?
There's no fixed timeline; it depends on how much material you have ready to upload. The mechanical steps are the same regardless of speed: create the world, add ICP and brand voice, add products and competitors, upload any knowledge base documents, and agents can start working the same day the core inputs are in.
Do I need a separate account for each client brand?
No, you don't need a separate account, you need a separate world within your existing plan. Founder includes 1 world, Team includes 5, Agency includes 25, and Enterprise is uncapped and custom.
Can a client see draft work before it publishes?
Yes, through the Dock, the client-facing review surface. Clients see what's waiting, mark it up with pins on the draft, send it back for revisions, and approve finished work into a Library.
Does a client's Dock access expose our internal notes?
No. A client seat is Dock-only, and internal team notes are never visible to a client, so your working comments and their review experience are kept separate.
If I onboard a second brand, does it start from zero like the first one did?
Each brand gets its own world, so a second brand's world starts empty in the sense that it doesn't inherit the first brand's ICP, voice or competitors. What it doesn't lose is the setup pattern: the same five-input sequence, the same playbooks and schedule templates, apply immediately, so the process is faster the second time even though the content of the world itself is separate.