The compiler for your business documents
Write presentations and documents as plain text — for content generated by humans, systems, and AI agents. Compile to a versioned semantic AST.
slidelang
A plain-text DSL for presentations. Compiles to HTML, PDF, and (soon) PPTX.
Explore slidelang →Why an AST-backed format
The DSL is one authoring interface — a human- and LLM-friendly way in. The real contract is the versioned AST underneath it: a stable, semver'd intermediate representation that any tool, agent, or renderer can consume. Today an AI writes.slidelang/.doclang source; tomorrow it can emit validated AST directly, or a third-party tool can produce either format from JSON, a spreadsheet, or a CRM.
Try it now
No install. Edit source, see the compiled output update live, right in your browser.
Open the playground →Built for agents
An MCP server, an LLM kit with calibrated prompts, and a machine-readable entry point.
llms.txt →Two Modes: Flexible and Strict
Because ZiraDocs serves both the messy drafting phase and the rigorous publishing phase, the compiler operates in two flavors:
Flexible Mode
Permissive parsing designed for AI agents and rapid human drafting. It tries its best to render your intent without failing on minor syntax omissions.
Strict Mode
Unforgiving validation designed for CI/CD and final compliance. If the AST doesn't perfectly match the schema, it fails the build.
Use Cases
Drafting with ChatGPT or Claude
(Flexible Mode) Hook our MCP server into Claude Desktop, or paste our llms.txt into ChatGPT. Have the AI draft entire decks or documents from your notes. Flexible mode ensures the agent's output renders gracefully even if the syntax isn't flawless.
System-to-Document API
(Strict Mode) Generate reports, pitch decks, or invoices directly from your database, CRM, or internal APIs. Emit the validated JSON AST natively and compile to PDF or PPTX in your automated pipelines with absolute structural certainty.
Pharma & Compliance (MLR)
(Strict Mode) For highly regulated industries. Enforce Medical, Legal, and Regulatory (MLR) rules in CI/CD. Ensure every claim is traceable, every document meets accessibility standards (PDF-UA), and block unapproved AI hallucinations from ever leaving your org.