We open-sourced a JSON-LD schema for brand identity to help bring your brand into the agentic world

When we built Markolé, we solved a specific problem: brand drift in AI-generated content.

Inside Markole, every document, every image, every marketing asset is generated with full access to your brand DNA, the narratives, the visual system, the positioning, the tone. It's not pasted into a prompt. It's deeply integrated. The output is on-brand by design, not by luck.

That's the whole point of the platform. You build your brand identity through AI-guided interviews, we synthesize it into structured DNA, and everything generated from that point forward is guaranteed to be aligned.

But here's the reality: people use other tools too. ChatGPT for a quick draft. Claude for brainstorming. Midjourney for visuals. An agency with their own stack. And the moment your brand leaves Markole as a PDF, all that structure is gone. The AI on the other end has to guess.

So we asked ourselves: how do we bring that same level of clarity and consistency to tools we don't control?

---

The problem with PDFs and copy-paste

When you hand a brand guide to an LLM, any LLM, it has to:

1. Parse the document (what's a heading? what's a caption? what's the actual rule?)
2. Interpret what each part means (is this the mission or the vision? core values or aspirational ones?)
3. Prioritize what matters for the task (which guidelines are critical vs. nice-to-have?)

Every step is a place where meaning gets lost. The tone drifts. The values get flattened. The colors come back approximate. And every person who prompts the AI describes the brand slightly differently, the drift compounds.

Inside Markole, none of this happens because the AI already has the structured brand DNA. But outside Markole? It's the wild west.

---

What we built: a machine-readable brand format

We created a JSON-LD export that takes everything Markolé knows about your brand and packages it in a format any AI tool can understand natively.

It's not a simplified summary. It's the real thing, semantically labeled so the AI doesn't have to interpret anything:

• Which text is the vision statement vs. the elevator pitch
• What the exact hex values are for light and dark mode
• Which font is for headings vs. body text
• What the tone of voice rules are, distinct from the personality
• Who the target personas are, with their goals and pain points
• The dos and don'ts for color, typography, imagery, and layout

We built it on JSON-LD with Schema.org as the base vocabulary, and published the schema definition openly at markole.com/schema/brand-dna.

---

How to use it

Inside Markole, you don't need it, the platform already has full brand context for everything it generates.

Outside Markole, this is where it changes things:

• Drop the .json into ChatGPT, Claude, or Gemini, the AI gets your complete brand identity in one file. No 500-word system prompt, no copy-pasting fragments from a PDF.
• Wire it into automated content pipelines, the JSON becomes your brand config. Every generated piece inherits the right tone, messaging, and visual references.
• Hand it to agencies and freelancers, instead of "please read the brand guide first," give them a file their AI tools can ingest directly.
• Use it in multi-agent workflows, when multiple AI agents collaborate on content, they all reference the same structured source of truth.

Will it be as good as generating directly inside Markole, where the brand DNA is deeply integrated? Honestly, no, but it gets significantly closer than a PDF or a pasted paragraph ever will.

---

What's in the export

Three scopes, depending on what you need:

Brand Report: All 17 narratives (vision, mission, values, archetype, personality, tone of voice, positioning, promise, story, manifesto, pitch, keywords, rituals, target audience), 4 meta-narratives, and the full brand summary.

Visual Identity: Color palettes (light + dark), typography, logo URLs in all variants, visual mood, concept statement, imagery guidelines, layout principles, and categorized dos & don'ts.

Brand Book: Both combined.

---

Why we made the schema open

We could have kept this proprietary. But brand identity shouldn't be locked into one platform's format, that defeats the purpose.

The schema at markole.com/schema/brand-dna is openly documented. Anyone can use the vocabulary to describe brand identity in JSON-LD, whether they use Markolé or not. We think structured, machine-readable brand identity is going to become standard infrastructure as AI takes over more of the content layer.

We built this for ourselves first, to make sure our users' brands stay consistent even when they leave the platform. Now we're sharing it because this is the kind of thing that gets better when more tools speak the same language.

If you're dealing with brand drift in AI-generated content, give the export a try on Markolé, or just use the schema in your own stack. The documentation is all there.