Tessryx

AI doesn't speak your CMS.
It speaks ours.

Your CMS was built for humans. Tessryx is built for both - content, workflows, and schemas all in the language LLMs natively understand.

Get early access

Zero config

Paste a JSON Schema.
Get a working CMS.

blog-post.schema.json

{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "BlogPost",
  "type": "object",
  "required": ["title", "body"],
  "properties": {
    "title":     { "type": "string", "maxLength": 120 },
    "slug":      { "type": "string", "pattern": "^[a-z0-9-]+$" },
    "body":      { "type": "string", "contentMediaType": "text/markdown" },
    "tags":      { "type": "array", "items": { "type": "string" } },
    "publishAt": { "type": "string", "format": "date-time" }
  }
}

Becomes a working content editor in Tessryx. No DSL. No configuration. No translation layer.

Interactive demo coming soon.

Tessryx is a content and workflow platform where everything - your schemas, your content, your automations - is JSON Schema. That means your agents can author it natively, your team can edit it through generated forms, and nothing is locked in a proprietary DSL. Paste a schema, get a CMS. Hand the same schema to an LLM, it knows exactly what to produce.

Content your agents can author.

Every schema is standard JSON Schema. LLMs can read it, validate against it, and produce content that fits - no translation layer, no custom SDK.

Workflows in the same language.

Automations are JSON documents validated against Tessryx's workflow schema. An agent can write a workflow the same way it writes content - via structured output.

Humans stay in the loop.

Generated authoring forms for every schema. Review agent output before it publishes. Provenance at the field level.

Native authoring

Agents as first-class authors.

Because schemas are JSON Schema and workflows are JSON documents, agents can build both natively. Hand an LLM a prompt like this:

create a product catalog schema with an AI-fill-description button

...and it produces two valid documents Tessryx runs immediately. No custom code, no plugin authoring, no SDK.

Full agent demo coming soon.

Why now

Why we're building this.

Every CMS was designed before AI agents were authors. They modeled schemas for humans and validators, not for language models. In the last two years that changed - agents are now producing structured content at scale, and the mismatch between what they speak (JSON Schema) and what CMSes speak (proprietary DSLs) shows up as glue code, retry loops, and fragile pipelines. Tessryx is what a content platform looks like when you design it for the way AI actually works.

Where it fits

The honest comparison.

Sanity, Contentful, Payload

Great content platforms. They weren't designed for AI-first workflows.

Directus

A database admin panel that grew into a CMS. Different tool for different jobs.

Tessryx

Built from the ground up for teams where agents and humans author the same content.

AI doesn't speak your CMS.It speaks ours.

Paste a JSON Schema.Get a working CMS.