Problem

Chunkr solves the one-size-fits-all problem in document parsing for RAG and LLM applications. PDFs, scans, spreadsheets, and office files often contain complex layouts, tables, formulas, handwriting, charts, and mixed reading orders that ordinary OCR or generic parsers cannot reliably turn into usable, citation-grounded data. AI teams consequently face a choice between accepting poor retrieval and extraction quality or building and operating a bespoke ingestion pipeline, creating engineering overhead and ongoing quality-control costs.

The central use case is document-grounded AI: converting financial reports, invoices, contracts, research papers, and other business records into dependable inputs for RAG systems, agents, and document-first applications. In operational workflows, the same capability can reduce manual data entry and errors in finance, legal review, compliance, and supply-chain processing.

Product / Service

Chunkr is a document-intelligence API with an open-source, self-hostable pipeline and a fully managed cloud service. It accepts PDFs, images, Word documents, presentations, and spreadsheets, then applies layout analysis, OCR, reading-order detection, semantic chunking, and vision-language-model processing to produce LLM-ready HTML, Markdown, or structured JSON. Its extraction layer supports custom schemas, citations, bounding boxes, and confidence scores, while teams can configure processing differently for ordinary text, tables, formulas, charts, or images.

The delivery model is designed to span experimentation through regulated production use. Developers can use the AGPL-licensed open-source version locally, while the Cloud API provides proprietary in-house models, managed infrastructure, and paid credit-based plans; enterprise customers can obtain VPC or on-premises deployment. The product’s benefit is control without having to build the entire ingestion stack: users can trade off speed, quality, and features while retaining document provenance and support for high-volume workflows.

Market

Chunkr competes in the document-parsing and document-intelligence market, increasingly positioned as infrastructure for AI ingestion, RAG, and automated document workflows. The category includes semantic parsing APIs such as LlamaParse and LandingAI; open-source foundations such as Docling and PyMuPDF; cloud OCR and document services from AWS, Google, and Microsoft; and enterprise automation products such as UiPath IXP. Chunkr’s distinctive angle is the combination of segment-level configurability, citation-aware outputs, self-hosting, and a managed API rather than a one-size-fits-all OCR endpoint.

The company has evidence of developer traction but limited publicly disclosed commercial scale. Its API and self-hostable pipeline launched in October 2024, reportedly generating 400,000 impressions, and by November 2024 had processed documents from healthcare, finance, research, government, education, and hardware users. The GitHub repository reports 4,044 stars and 264 forks, while the commercial site offers paid plans from $375 to $2,000 per month plus enterprise pricing. PitchBook lists a $500,000 financing amount and a generating-revenue status, although it does not disclose a current revenue figure or verified paying-customer count, so the business should be viewed as early commercial traction rather than a proven scaled vendor.