Ragie

Scanned checks are one of the hardest documents to parse reliably. Handwriting, inconsistent layouts, security patterns, occasional upside down pages. A customer used Ragie Parse on a batch of them last week. Parse is backed by Agentic OCR. It reads documents the way a human would, element by element, regardless of how messy the source material is. Learn more at ragie.ai/parse

It's tax day. Vibe code your tax assistant agent with Ragie’s new skill.  https://lnkd.in/g9WrJuzk The skill gives your model the context it needs to work with Ragie's SDK Install it in seconds: npx skills add ragieai/skills To show what's possible (and stay on theme), we built a Tax Document Assistant with it: a CLI that ingests your tax docs and IRS publications and answers natural language questions with citations back to the source file. Full tutorial: https://lnkd.in/gyayS_9R

A few reliability improvements shipped this week. HTML, EPUB, and Excel files all handle edge cases more reliably now, password-protected PDFs are caught immediately with a clear response, and documents paused due to insufficient credits resume automatically once credits are added. Stay up to date: ragie.ai/changelog

Ragie Parse now supports Agentic OCR, available in beta, and it unlocks a new class of document extraction. Standard parsing works well for clean, structured documents. But a lot of real-world documents aren't clean. They have forms, tables, stamps, signatures, handwritten notes, logos, and inconsistent layouts. Agentic OCR handles all of it, using vision models to extract these as structured elements you can access directly through the Elements API. The highest accuracy option for visually complex documents. Docs: https://lnkd.in/g-dDvEzM Stay up to date: ragie.ai/changelog

Ragie's Document Elements API is now available, and it's one of our most requested features. The new GET /documents/{document_id}/elements endpoint gives you direct access to every element extracted from a document in reading order: titles, tables, images, code blocks, and more. Each element includes its type, text, markdown, page location, and bounding box, with type-specific content where relevant. Filter by element type or index range to target exactly what you need. Stay up to date: ragie.ai/changelog

Ragie Instructions now support Context Templates. Instructions are Ragie's way of telling the platform what structured data to extract from your documents. They now support an optional context template that prepends document metadata to each extraction call. This gives you fine-grained control over what context the model sees before processing each chunk, leading to more accurate results across complex, multi-document workflows. Stay up to date: ragie.ai/changelog

Ragie Parse now supports Form & Signature Extraction. Ragie's new agentic OCR pipeline can capture handwritten signatures, detect whether they're signed or unsigned, classify form fields by type, and extract key-value pairs from label-dense documents like invoices and contracts. If you're building agents and apps that process real-world documents, this is the kind of structure that makes the difference. Learn more: ragie.ai/parse Stay up to date: ragie.ai/changelog

Introducing Ragie Parse, now in early access. Traditional OCR gives you a wall of text and loses all the structure. Ragie Parse uses a new Agentic OCR pipeline to extract structured content with higher fidelity — supporting 25+ element types including tables, forms, signatures, key-value pairs, barcodes, and stamps. If you're building agents and apps, better parsed documents mean better context. It's that simple. Learn more: ragie.ai/parse Stay up to date: ragie.ai/changelog

We've improved spreadsheet extraction in Ragie. Spreadsheets aren't just tables — they're often full of embedded images, charts, and freeform content that traditional extraction misses entirely. Ragie now extracts all of it alongside your table data, giving your RAG pipeline a more complete picture of what's actually in the document. No configuration needed. It just works. 👉 ragie.ai/changelog

PowerPoint support, improved — PPTX and PPT files now benefit from the same high-resolution extraction pipeline as native PDFs. Stay up to date → https://lnkd.in/g7ktMea2