Documents in. Markdown or JSON out.
Parse complete documents into page-aware Markdown, or extract the exact JSON fields your application needs. Run both workflows locally from the UI, CLI, or REST API.
Local-first document processing · Developer preview
01 · Parse
Keep the complete document
Turn PDFs and images into ordered, page-aware Markdown.
02 · Extract
Return only the fields you need
Define a JSON Schema and accept only results that satisfy it.
03 · Integrate
Build on one stable contract
Drive the same resources through the Web UI, CLI, or OpenAPI 3.1 API.
Choose your path
Section titled “Choose your path”01
Parse a document to Markdown
Use the prebuilt parser and inspect complete-document and per-page output.
02
Extract a document to structured JSON
Run a known receipt through a schema and inspect the validated result.
03
Connect your model provider
Use the bundled NuExtract3 runtime, OpenAI, Microsoft Foundry, Ollama, or another OpenAI-compatible endpoint.
04
Integrate the REST API
Use generated endpoint documentation, exact schemas, and copyable requests.
Two workflows. One local stack.
Parsing preserves a document as Markdown. Extraction selects fields into JSON. You can use either workflow by itself or use both for the same source.