Your brand in one canonical file. Owned by you, signed, and updated in one place. Every tool you authorize reads the current version before it writes in your name.
Encode once. Control everywhere.
--- bcp_version: "0.8" file_type: pointer brand_name: "Encoded Brands" last_updated: 2026-09-20 publication_profile: registry_backed canonical_bcp: https://registry.brandcontextprotocol.dev/encoded-f2fa3140/.well-known/brand.md registry_handle: encoded-f2fa3140 registry_mcp: https://registry.brandcontextprotocol.dev/mcp?customer_handle=encoded-f2fa3140 agent_first_action: "fetch https://registry.brandcontextprotocol.dev/encoded-f2fa3140/.well-known/brand.md" ---
Copy one short starter into a new chat. It points your AI to Spark, uses the source material your chat can access, and shows you a draft for confirmation before any publish step.
Help me encode my brand into a public Brand Context Protocol. Start here: https://encoder.encodedbrands.ai/.well-known/agent-commerce.json. Use the owner-authorized spark/MCP handoff in that discovery document, not the Encoder GUI. Show me the proposed result for confirmation. Do not publish until the owner approves.
The Brand Context Protocol (BCP) turns everything your teams already know into one structured, agent-readable source. Every department, every tool, every freelancer and agency you work with reads the same brief instead of inventing their own.
Your voice, your claims, your boundaries, held in one source you own. Read once, reused everywhere, always the current you.
Every team and tool that touches your brand reads from one place you manage, not twelve people improvising.
Brand context gets re-typed into every prompt, every tool, every time. A canonical source is read once and reused.
Without your voice on record, AI fills the gap with the average of everyone else in your category.
Start free, upload what you have, and participate in a focused interview. The Encoder extracts the signal and writes your first BCP file tree. The questions can be pretty deep. If you need to come back to one later, go ahead and skip it. Your completed BCP remains available in Registry while you keep it published, and you can withdraw it.
Your brand context becomes a signed endpoint hosted in Registry by default. Agents and tools you connect can fetch it before they write, classify, sell, support, or summarize. The record remains available while you keep it published, and you can withdraw it. Paid plans add teammates and fresh AI-assisted re-encoding; manual download, edit, upload, backup, and republish remain available without using an encode.
The result: your BCP, signed and available to the agents you connect.
Your BCP is a living document, not a one-time file. Edit it any time on encodedbrands.ai. Once you save, that update is ready for every AI and agent already connected to your BCP the next time they read it, so you never have to repeat the update on each surface by hand.
Your first full encode and signed Registry record are free. The record remains available while you keep it published, and you can withdraw it. Every Registry publication is integrity-signed, and DNS domain verification is free. Upgrade when you need to control access, know who used gated material, and prove what was live.
Start with one, free, complete encoding. Get your BCP files and a signed Registry record that stays available while published and can be withdrawn. Verify your domain at no charge, revise manually for free, or re-encode using the Encoder for $49. The free tier includes no human support. Docs, community, and self-serve tooling only. Human support is a Certified feature.
Start your free encode →Registry reads, integrity signing, and optional DNS verification are free. Registry Certified adds named audit and access control for authenticated gated material, certification status, team access, and support. Read the docs →
Brand Context Protocol creates a standing brief that AI tools and agents you connect can read before they write, place, or represent your brand. The result can be more consistent, more faithful output without re-briefing each connected tool every time.
AI tools do not reliably start with your current brand context. Many teams paste guidelines into system prompts one tool at a time, which consumes session context and can drift. BCP gives the agents you connect access to the same brief, from the same source, updated in one place.
Your brand guidelines matter, but PDFs are not a reliable machine-readable source across AI tools. Encoded translates your guidelines into a structured form that tools you explicitly connect can read, subject to each tool's URL or MCP support and controls.
Encode, publish, then update. Start in the AI your team already uses with Spark or MCP, or use the guided Encoder. Either path encodes the brand you already have. You review, then publish. Your BCP is hosted in Registry by default after you publish. Update anytime. Connected agents pick up the current version the next time they read it. Start with Spark: copy the starter on this page, or point your agent at https://encoder.encodedbrands.ai/.well-known/agent-commerce.json. Prefer the guided Encoder? Open https://encoder.encodedbrands.ai/encode. Connect compatible tools with the MCP guide and docs. Those are custom remote connectors you configure, not a native endorsement by the tool vendors. Your web team can also serve /.well-known/brand.md on your own domain as a short pointer to that Registry record.
Your brand voice, claims, and boundaries are designed to be read by agents. That's the point. You control what goes in. Nothing is published without your review. The Registry hosts your BCP files until you take them down. Gated access to confidential messaging matrixes is coming soon.
The Encoder session takes 30 minutes to an hour. Enterprise engagements are scoped to your organisation, typically one to two weeks from kickoff to published BCP.
Whatever you have. Existing brand guidelines, tone-of-voice docs, positioning decks, campaign examples. The Encoder works with what exists. If you have nothing, the interview builds your BCP from scratch.
By default, your published BCP is hosted at a Registry URL. Serving it at /.well-known/brand.md on your own domain requires setup by you or your web team. The recommended domain file is a short pointer to the Registry record, not a copy of the full package. Once configured, tools with URL access can fetch it there. Agents that search Registry can also discover your published record without already knowing your domain.
Your team, your agencies, and the AI tools you already use can work from the same brief, once you connect them. Public Registry reads are free over a URL or Model Context Protocol (MCP). You can add Encoded's public Registry MCP as a custom remote connector in ChatGPT, Claude, and Grok, subject to each product's plan, client, and administrator controls. Agencies and providers can use the same source where their tools permit URL or MCP access. Need gated private context? Registry Certified adds access control, team governance, version history, and audit evidence for who pulled protected context, when, under which grant, and against which section and content hashes. You configure those connections. It does not mean those vendors natively adopt, automatically read, or endorse BCP or Encoded Brands. Setup notes live in the MCP guide.
We use it ourselves. Our brand briefing is live at encodedbrands.ai/.well-known/brand.md. The same file our own agents read before they write anything under our name. If we wouldn't use it for our own brand, it doesn't ship.
A company that makes AI sound like your brand had better sound like itself. So ours is live and public at encodedbrands.ai/.well-known/brand.md, the same brief our own agents read before they write a word under our name.
Whatever voice, structure, and decisions we bake in carry across every brand that builds on the standard. We set the example in public, with our own name on it.
If we wouldn't use it for our own brand, it doesn't ship.
--- bcp_version: "0.8" file_type: voice parent: https://registry.brandcontextprotocol.dev/encoded-f2fa3140/.well-known/brand.md last_updated: 2026-08-05 --- # Voice We sound like a senior brand strategist who has shipped the work and read the spec. Plainspoken with marketers. Precise with engineers. Honest about what ships today and what is on the roadmap. We frame in marketer language and prove in technical language, never the other way around. ## Personality - **Plainspoken.** We say what the product does in words a CMO uses. We do not hide behind category jargon. Not blunt for its own sake; never burying the point in qualifications. - **Honest about status.** We publish gaps as gaps. "In development" beats silence beats overclaim. We launch the narrow case as the narrow case. - **Committed to the bit.** When we do voice, we commit. No winking, no hedging. The ACME example in our docs does not break character; neither do we. - **Quietly confident.** We raise the floor under marketers afraid of AI slop. We do not promise the ceiling. ## Register The register shifts by surface, not by mood. Marketer register is the default for public-facing surfaces. Tech register is the default for the spec, the docs, and the files themselves. Within a single surface we code-switch when the topic demands proof, then return to the framing register. ## Vocabulary ```yaml prefer: - "making your AI and agentic tools sound like your brand" - "encode your brand once, use it everywhere, update as often as you like" - "the files are the product" - "better content, fewer rounds of revisions, faster speed to quality output" - "voice modulation by audience or market" - "living document" - "open spec" - "agent-readable" - "much more than" (when an accurate contrast is needed) - "in development" (for roadmap features) - "International Orange" (for the brand color) avoid: - "perfect" (we raise the floor, not the ceiling) - "guaranteed" (in marketing claims; permitted only in technical guarantees) - "revolutionize" (category-generic) - "leverage" (category-generic) - "unlock" (category-generic) - "Brand OS" (not us, and we reject the framing) - "brand DNA" (generic AI-startup mush on marketer surfaces) - "our tool" / "our platform generates" (we are a protocol, not a tool) - "not just X" when we in fact do not do X (misstates the product) - "trusted by [vendor]" without a signed paper - moon-and-stars promises ``` ## Punctuation and formatting rules ```yaml oxford_comma: yes headline_periods: yes exclamation_points: never em_dashes: never sentence_case_headlines: yes capitalization_notes: "Brand Context Protocol and BCP are capitalized. International Orange is capitalized. File paths and frontmatter keys appear in code voice." ``` ## Channel modulation ```yaml contexts: index_page: tone: "plainspoken, outcome-led, marketer register" register: "casual-professional, no jargon" example: "Give the AI & agents you connect the same source of brand truth." docs: tone: "precise, technical, complete" register: "developer-facing, frontmatter and paths welcome" example: "Serve markdown as text/plain; charset=utf-8 with Access-Control-Allow-Origin: *." linkedin: tone: "marketer register opens, code-switches to tech for proof, returns to marketer for the close" register: "mixed audience, frame-then-prove" example: "Your agents are writing copy this quarter. Here is the file that tells them how you sound." conference_stage: tone: "speak to the room's center of gravity with explicit nods to the other half" register: "named code-switch, not smuggled" example: "For the implementers in the room, here is what that means in the spec." sales_deck: tone: "different sections, different registers" register: "framing sections marketer, technical sections tech, single deck" example: "Section 1 talks outcomes. Section 4 shows the file tree." one_to_one_email: tone: "match the recipient" register: "marketer to marketers, technical to engineers" example: "Hi Priya, here is how the Registry handles signing." ``` ## Code-switch rule Frame in marketer register. Prove in tech register. Never invert. Do not open in tech register and then translate down for the marketer in the room. Do not open in marketer register and refuse to drop into spec when asked. Inversions feel like condescension in both directions. ## Audience hierarchy Agents read first. Humans read second. The BCP files themselves are written for agents to consume. The marketing site speaks to humans about what their agents will do. ## Copy examples | On-brand | Off-brand | |----------|-----------| | Give the AI & agents you connect the same source of brand truth. | Revolutionize your brand with AI-powered brand intelligence. | | Encode your brand once. Use it everywhere. Update as often as you like. | Unlock your brand DNA across the agentic economy. | | Drift detection is in development. | Our Brand OS guarantees perfect agentic content. | | Much more than your logo and colour palette. | Not just another brand guidelines tool. | | A brand like Acme could publish a BCP that... | Trusted by Acme, OpenAI, and Anthropic. |
Questions about encoding your brand, Registry Certified, or Enterprise? Send us a note. We will reply by email.