How it works
You own the brand. It shouldn’t route through you.
One loop, four steps: import your brand, structure it, publish it, and connect it to the tools your team already writes with. Set the standard once — the system holds it after that.
Import
You: Upload your brand PDF or paste your website URL.
The system: AI extracts the system — colours, type, logos, voice, messaging — into a structured draft in about ten minutes. You review, correct, and confirm. Nothing is published until you say so.
The document you keep re-explaining becomes a model you can point at.
Structure
You: Refine the parts only you know — the messaging house, tone with do/don't pairs, boilerplates, approved and banned phrases.
The system: Holds the visual and the verbal identity as structured, versioned data — not page content. Every later change is one edit, tracked.
The verbal half of your brand finally exists outside your head.
Publish
You: One click.
The system: A clean public brand page at one URL, plus the logo pack — always the current version. Share it with a new hire, an agency, a partner.
“Where’s the latest logo / boilerplate / deck template” stops being a message you receive.
Connect
You: Add your brand's endpoint to Claude, ChatGPT or Cursor — once.
The system: Your published brand is a live MCP server. Every connected tool reads your voice rules, boilerplates and banned phrases before it writes.
Drafts start on-brand. You review exceptions, not everything.
One model, two audiences
Publishing serves your people. Connecting serves their tools.
Both read the same brand model, so they can never drift apart — update once and everyone, human or AI, works from the new version.
For your people
Public brand page
One URL, always current. No logins, no stale PDFs.
Share links
Hand an agency or partner the current version — not an attachment.
Logo pack
Every variant, correct formats, one download — straight from the page.
For their tools
https://app.spokepros.com/api/mcp/your-brandYour published brand is a live MCP server — Claude, ChatGPT and Cursor read your voice rules, boilerplates and banned phrases before they write. The same model is served as llms.txt and brand.json for anything that reads a URL.
Once it’s live, the work compounds.
Everything downstream reads the one model, so nothing contradicts you: the Brand Assistant drafts posts, emails and talking points in your voice. Brand check scores any draft — or a live URL — with specific fixes. Performance analysis ties your content results back to your own pillars. And version history keeps one current truth as the brand evolves.
Fair questions.
How long does the import really take?
About ten minutes for the AI pass, then however long you spend reviewing. Most brands are published the same sitting. Nothing goes live until you confirm it.
What if the AI gets something wrong?
You review and correct every section before publishing — the import is a head start, not an authority. On paid plans you can re-run it as your brand evolves.
What do my teammates actually see?
A clean brand page at one URL — no logins, no downloads. And if they write with Claude, ChatGPT or Cursor, their tool reads the same model underneath.
Do I need my design team to set this up?
No. If a brand PDF or a website exists, that's enough to start. Design keeps living in your design tools — SPOKEPROS holds the system that keeps everyone honest.
What does it cost to try?
Nothing. The Free plan includes the import, the published page, and the brand server. Paid plans add room — more editors, more generations, brand check.
Run the loop once. Keep the afternoon.
Import, structure, publish, connect — on the Free plan, in one sitting.