Skip to content
● New: team templates. Start a loop from the gallery and decide round by round. See how it works →
Brand guardrails for AI agents

Your copy is written in your repo. Your brand belongs there too.

Give every AI agent your brand memory, and a number that proves it landed.

Marketing copy, email texts, and UI microcopy are increasingly written in the editor, by an agent that doesn't know your brand. With Branddock in Cursor, that agent pulls in your brand voice and brand rules before it writes, and you can check an output before you merge it. Your BRAND.md can also just sit in your repo root.

Connect in 30 seconds

  1. Add `branddock.app/mcp` as an MCP server in your Cursor settings.
  2. Authorise and select your brand.
  3. Ask for copy, or have an existing string scored against your brand.

Works in any MCP-compatible editor, such as Cursor, Claude Code, and Windsurf.

Try it right away

Paste this prompt as soon as the connector is set up. You'll get text in your brand voice, with a brand score underneath.

Rewrite the hero copy in `page.tsx` in our brand voice and score the result. Show the old and new versions side by side.

Everyone can write. We show the number.

The brand score tests your copy against your brand, from 0 to 100, with three checks: Brand style, Strategy and Human. For each finding you see what jars and why. Free, also via the API. No more debate about whether something is on-brand: there is a number.

How the brand score works

Running multiple client brands?

On Agency you monitor 15 brands for €349 per month, roughly €23 per client brand. Each brand keeps its own brand DNA, its own voice, and its own score, and your agent works in the brand you select.

Consumer: €422.29 incl. 21% VAT

See the agency setup

28 days free · 300 credits · no credit card