Reducto

Reducto is an image analysis tool. Reducto parses, extracts, splits, and fills documents using its r-1 model, with layout-aware outputs, bounding-box citations, and enterprise-grade compliance for AI pipelines.

Reducto is an agentic document platform built for teams that need to turn messy PDFs, scanned forms, and long reports into clean, structured data. The core of the product is a parsing model called r-1, which reads layout, text, tables, figures, and formatting together in context. It returns document structure with bounding-box citations, so every extracted value can be traced back to its source on the page. The platform also includes dedicated tools for extraction, splitting, and editing, which together cover the common pipeline stages that usually require stitching multiple services. Extract pulls schema-defined fields from documents such as invoices, onboarding forms, and financial disclosures. Split separates multi-document files or long forms into individually useful units using layout-aware heuristics, which removes manual pre-processing. Edit fills in detected blanks, tables, and checkboxes with supplied data, without needing predefined templates or bounding boxes. Additional features include intelligent chunking, figure summarization, graph extraction, automatic page rotation, and embedding optimization, all aimed at making document data ready for downstream language model workflows. Reducto is aimed at AI engineering teams and enterprises in finance, healthcare, insurance, and legal, as well as startups building document-heavy applications. Its public references include Harvey, Scale AI, Vanta, Legora, and JLL, which suggests a strong presence in legal and financial document processing. The platform is positioned as production infrastructure rather than an end-user app, with claims of 99.9%+ uptime, SOC2 and HIPAA compliance, and the option to deploy inside your own environment for strict data residency requirements. Among alternatives, Reducto differentiates by combining parsing, extraction, splitting, and form filling in one platform, with a single model (r-1) replacing multi-tool pipelines. The r-1 model is priced at 1 cent per page, all-in, and is available in preview. Teams that want a managed API for document intelligence with strong table handling and citation-level grounding will find it a close fit; those seeking a lightweight browser tool for occasional PDF conversion may find it heavier than needed.