# Reducto — Full Overview for LLMs

> Reducto is the complete agentic document platform for leading AI teams that need performance at enterprise scale. Document work starts here: the most accurate document processing available — parsing, extraction, splitting, and classification — plus document editing, workflows, and agent-ready tooling, in one platform.

This file is the extended companion to https://reducto.ai/llms.txt. For complete, always-current API documentation, prefer https://docs.reducto.ai/llms.txt — every docs page there is available as raw Markdown by appending `.md` to its URL.

## Overview

Reducto provides a comprehensive toolkit for working with documents the way a human would, combining custom in-house models with leading frontier models to power efficient, accurate document workflows. It is used by AI-native companies and document-heavy enterprises — teams at Harvey, Scale AI, Vanta, and organizations up to Fortune 10 scale. Over 5 billion pages processed and counting.

Three things define the platform:

1. **Performance for you** — zero-shot accuracy on complex documents where other solutions aren't production ready, with automatic routing across 12+ orchestrated models to balance accuracy, latency, and throughput. Handles the long tail — tables, charts, figures, handwriting, scans — without templates or retraining. Deep Extract achieves 99% recall and precision on micro1's LongExtractionBench.
2. **Enterprise readiness** — deployable anywhere (hosted, VPC, on-premises, air-gapped), SOC 2 and HIPAA compliant with zero data retention by default, autoscaling for bursty workloads, white-glove support, and custom SLAs.
3. **Complete toolkit** — every document task in one platform: parse, split, extract, classify, and edit across 30+ file types, from raw file ingestion through agent-ready outputs and workflow orchestration.

**Get started free:** sign up at https://studio.reducto.ai/ for 15,000 free credits. AI agents can begin with the Agent skill at https://reducto.ai/SKILL.md, then connect through the MCP server or CLI.

## Products

- Parse — the ingestion layer: PDFs, scans, spreadsheets, and slides into structured, citation-ready JSON. https://reducto.ai/parse · https://docs.reducto.ai/api-reference/parse.md
- Split — the segmentation layer: describe sections in plain language, get page ranges back. https://reducto.ai/split · https://docs.reducto.ai/api-reference/split.md
- Extract — the extraction layer: typed JSON with a citation and bounding box on every value. https://reducto.ai/extract · https://docs.reducto.ai/api-reference/extract.md
- Classify — the routing layer: plain-language taxonomy, best match plus per-criterion confidence. https://reducto.ai/classify · https://docs.reducto.ai/api-reference/classify.md
- Edit — closes the loop: write extracted data back into finished PDF and DOCX files. https://reducto.ai/edit · https://docs.reducto.ai/api-reference/edit.md
- Studio — the no-code workspace, same engine as the API, included with every plan. https://reducto.ai/studio · https://studio.reducto.ai/

## Developer Tools

- Agent skill: self-contained setup instructions for AI agents, including API usage, CLI, and MCP server instructions. https://reducto.ai/SKILL.md
- MCP server: hosted at https://mcp.reducto.ai/mcp (Bearer API key) or local via `uvx mcp-server-reducto` — guide at https://docs.reducto.ai/mcp-server, server cards at https://reducto.ai/.well-known/ai-catalog.json
- CLI: `pip install reducto-cli` — parse, extract, split, classify, and edit from the terminal. https://docs.reducto.ai/cli
- Agent integration guide: https://docs.reducto.ai/agent-guide.md · SDKs (Python, Node): https://docs.reducto.ai

## Pricing, Security, Industries

- Pricing: credit-based Standard / Growth / Enterprise plans, 15,000 free credits at https://studio.reducto.ai/ — https://reducto.ai/pricing
- Security: SOC 2, HIPAA (BAAs on Growth+), zero data retention, deploy hosted/VPC/on-prem/air-gapped — https://docs.reducto.ai/security/policies
- Industries: finance, healthcare, legal, insurance, government, construction, logistics — https://reducto.ai/industries · case studies at https://reducto.ai/customers

## Comparisons

Honest tool-by-tool comparisons — accuracy, deployment, tooling, pricing, and when each option fits. Hub at https://reducto.ai/compare.

- [Reducto vs LlamaParse](https://reducto.ai/compare/reducto-vs-llamaparse): A complete agentic document platform with benchmark-leading extraction vs a developer-friendly parser in the LlamaIndex ecosystem.
- [Reducto vs Unstructured](https://reducto.ai/compare/reducto-vs-unstructured): A managed agentic document platform with measured accuracy vs an open-source document parsing and ETL library for LLM pipelines.
- [Reducto vs AWS Textract](https://reducto.ai/compare/reducto-vs-aws-textract): A complete agentic document platform that can run in your own VPC vs a cloud OCR primitive inside the AWS ecosystem.
- [Reducto vs Azure Document Intelligence](https://reducto.ai/compare/reducto-vs-azure-document-intelligence): A zero-shot agentic document platform with the full toolkit in one API vs a cloud OCR service with prebuilt and custom-trained models.
- [Reducto vs Google Document AI](https://reducto.ai/compare/reducto-vs-google-document-ai): A zero-shot agentic document platform that deploys anywhere vs cloud OCR processors you configure per document type.
- [Reducto vs Gemini](https://reducto.ai/compare/reducto-vs-gemini): An agentic document platform that orchestrates frontier models inside a production pipeline vs a general-purpose frontier LLM used raw for document work.
- [Reducto vs Extend](https://reducto.ai/compare/reducto-vs-extend): A complete agentic document platform with benchmark-leading extraction and flexible deployment vs a managed document workflow product.
- [Reducto vs Pulse](https://reducto.ai/compare/reducto-vs-pulse): A complete agentic document platform proven at enterprise scale vs a focused document parser with a financial-documents emphasis.
- [Reducto vs Datalab](https://reducto.ai/compare/reducto-vs-datalab): A complete agentic document platform proven at enterprise scale vs the research-driven vendor behind the open-source Marker and Surya models.
- [Reducto vs ABBYY](https://reducto.ai/compare/reducto-vs-abbyy): An AI-native agentic document platform with zero-shot accuracy vs decades-old enterprise OCR with template-based capture.

## Blog

Technical articles, case studies, and product announcements. This is an index — fetch each URL for the full article.

- [Built to Scale: How Doe Automated Over 8 Million Agent Tasks with Reducto](https://reducto.ai/blog/reducto-doe-ai-agent-customer-story) (2026-07-21): Doe has completed over 8 million tasks for customers through its AI agent command center, using Reducto to handle complex document work at scale.
- [Deep Extract vs. Frontier Models vs. Humans: The Real Tradeoffs for Long Structured Extraction](https://reducto.ai/blog/reducto-deep-extract-mode-vs-frontier-models-vs-humans) (2026-07-20): Deep Extract trades a higher cost for near perfect accuracy on long, high stakes documents, compared with the manual review process most teams still rely on.
- [Reducto Deep Extract Leads Benchmark on Complex Document Extraction](https://reducto.ai/blog/reducto-leads-benchmark-complex-document-extraction) (2026-06-30): micro1 released an independent benchmark evaluating document extraction systems on difficult, high-field-count workloads. Reducto Deep Extract ranked first overall.
- [Reducto Raises Frontier Model Accuracy on GDP.pdf](https://reducto.ai/blog/reducto-raises-frontier-model-accuracy) (2026-06-16): Frontier models score under 30% on GDP.pdf, a benchmark of 100 real professional documents. We ran the same tasks with Reducto's structured parse added: macro accuracy jumped 9pp, reasoning tokens dropped 13%, and answers arrived faster despite 82% more input.
- [What is an Agentic Document Platform?](https://reducto.ai/blog/reducto-what-is-an-agentic-document-platform) (2026-06-11): Learn about what makes an agentic document platform, and how it differentiates from previous IDP and OCR solutions.
- [Parsing the 10-K: why financial filings defeat standard PDF pipelines](https://reducto.ai/blog/10k-document) (2026-06-08): Most financial RAG pipelines fail before the model ever runs. Here's where the parser loses the structure and what it takes to get it back.
- [Build your first document workflow with Reducto](https://reducto.ai/blog/build-document-feature-with-reducto) (2026-06-08): Building a document feature or integrating Reducto into your existing document workflow should be quick and easy. We'll show you how.
- [Deep Split: Utilizing Agent Harnesses for Accuracy at Scale](https://reducto.ai/blog/reducto-deep-split-agent) (2026-06-03): We're introducing a new version of split for longer documents and workflows with large numbers of categories.
- [Announcing Reducto’s Classify Endpoint: Route Documents Before Processing](https://reducto.ai/blog/reducto-classify-endpoint-api) (2026-05-21): Learn about Reducto's new Classification endpoint, which helps categorize documents in a fast and lightweight way before downstream document work.
- [Reducto acquires Opennote](https://reducto.ai/blog/reducto-acquires-opennote) (2026-05-07): We're announcing our acquisition of Opennote, the AI notebook that helps students understand, organize, and practice from their notes.
- [Introducing Smart Schema](https://reducto.ai/blog/reducto-smart-schema-extract-optimization) (2026-04-28): Accurate extraction starts with your schema. Today, we're introducing Smart Schema in our Studio experience to help teams autonomously create contextually aware schemas and improve them automatically.
- [How Harvey Turned OCR Quality Into Customer Confidence with Reducto](https://reducto.ai/blog/reducto-harvey-legal-ai-customer-story) (2026-04-07): Read how Harvey partners with Reducto to upgrade its document processing capabilities, going from evaluation to full production in roughly six weeks while improving accuracy across complex legal documents like handwritten notes, redlines, and image-based files.
- [Introducing Deep Extract](https://reducto.ai/blog/reducto-deep-extract-agent) (2026-04-06): We're releasing our most powerful update to structured extraction yet with Deep Extract.
- [How Vanta is Building AI-Native Compliance with Reducto](https://reducto.ai/blog/vanta-reducto-case-study) (2026-02-20): Learn how Vanta's AI team utilize Reducto to power their AI-native compliance workflows, including questionnaire automation and evidence evaluation.
- [Reducto Is Now Available on AWS Marketplace](https://reducto.ai/blog/aws-marketplace-reducto) (2026-02-06): Reducto is now available on AWS Marketplace, enabling enterprises to purchase the leading AI document intelligence platform using their committed AWS spend.
- [Scaling Beyond Annotation: How Reducto Powers Scale AI’s Agentic Expansion](https://reducto.ai/blog/scale-ai-reducto-case-study) (2025-12-15): See how Scale's Public Sector and Enterprise teams use Reducto to power their new agentic workflows with high precision, security, and scale.
- [How Reducto is building SOTA chart extractions](https://reducto.ai/blog/reducto-chart-extraction) (2025-12-03): Our approach to developing a new chart extraction method, returning near pixel-perfect data accuracy for complex line graphs, bar graphs, and more.
- [August’s Competitive Edge: Legal AI Built on Reducto](https://reducto.ai/blog/august-law-case-study) (2025-11-24): August delivers attorney-ready legal workflows by pairing its agentic platform with Reducto’s high-fidelity document understanding, enabling accurate, traceable work products even from the hardest-to-parse documents.
- [How LEA Uses Reducto to Deliver Document Intelligence to Enterprise Wealth Management Firms](https://reducto.ai/blog/lea-reducto-case-study) (2025-11-11): LEA uses Reducto’s secure document AI to automatically organize, extract, and pipe data from thousands of complex financial documents—helping $10B+ RIAs scale operations, cut manual data entry in half, and grow without adding headcount.
- [Reducto Announces $108M in Funding to Define the Future of AI Document Intelligence](https://reducto.ai/blog/reducto-series-b-funding) (2025-10-14): We're announcing $108M in total funding following a new $75M Series B led by a16z. See what we've learned and our plans for the future.
- [How we did a database migration without logical replication - with zero downtime](https://reducto.ai/blog/reducto-database-migration-zero-downtime) (2025-09-18): Database migrations are tough - here's how we did it using Pgdog and database mirroring.
- [How Elysian uses Reducto to Review Insurance Claims 16x Faster](https://reducto.ai/blog/reducto-elysian-case-study) (2025-09-04): Read about how Elysian utilizes Reducto to process insurance claims of all shapes and sizes, resulting in a 16x faster audit process compared to traditional methods.
- [Enterprise RAG at scale: search techniques for million-document databases](https://reducto.ai/blog/reducto-ingestion-rag-enterprise-scale) (2025-08-26): How to build reliable retrieval systems on top of Reducto’s parsing and chunking pipelines—so your enterprise AI can handle massive unstructured document ingestion with accuracy.
- [Streamlining Document Processing with Reducto and Databricks](https://reducto.ai/blog/streamline-document-processing-reducto-databricks) (2025-06-26): How to unlock your unstructured data with Reducto, with Databricks.
- [How Gumloop Enables Anyone to Make AI Workflows with Reducto](https://reducto.ai/blog/gumloop-case-study) (2025-06-24): Read how Reducto powers Gumloop's AI automation of enterprise workflows, allowing non-technical teams at Instacart and Webflow to rapidly scale AI adoption.
- [How Benchmark uses Reducto to Build the First-Party Data Engine for $1 Trillion in Assets](https://reducto.ai/blog/benchmark-case-study) (2025-06-18): Benchmark is an AI-native investment platform used by firms managing ~$1T in assets under management, including some of the world’s top financial institutions. See how they utilize Reducto in some of their most important features, such as Document Builder.
- [How to use Reducto parsing with Elasticsearch for Semantic Search](https://reducto.ai/blog/how-to-reducto-parsing-elasticsearch-semantic-search) (2025-06-05): Demonstrating how Reducto's document processing API can be integrated with Elasticsearch for semantic search.
- [How Anterior Accelerates Prior Authorization and Clinical Decision-Making with 99%+ Accuracy with Reducto](https://reducto.ai/blog/anterior-case-study) (2025-06-05): Learn how Anterior accelerates prior authorizations with 99%+ precision using Reducto’s document ingestion engine—turning unstructured medical records into real-time decisions.
- [Automating Enterprise Workflows at Scale with Stack AI and Reducto](https://reducto.ai/blog/reducto-stack-ai-case-study) (2025-05-29): Learn how Stack AI customers have processed over 5M+ documents using Reducto in their enterprise automation workflows.
- [How to Use AI to Extract Data from Claim Submissions at Scale](https://reducto.ai/blog/extract-api-health-insurance-claims) (2025-05-09): Health insurance is a unique space that can benefit greatly from LLMs and proper document ingestion at scale - use AI to help automate extracting important data with accuracy.
- [Reducto raises $24.5M Series A to help enterprises unlock unstructured data](https://reducto.ai/blog/reducto-series-a-funding) (2025-04-25): We raised a $24.5M Series A led by Benchmark to help enterprises turn unstructured data into accurate, LLM-ready inputs—at scale. Read more on what we've built and where we're headed next.
- [Extraction Trouble? Here Are 5 Pitfalls to Avoid when Configuring Your JSON Schema](https://reducto.ai/blog/document-ai-extraction-schema-tips) (2025-04-16): If your extraction outputs aren't as expected, troubleshoot first by checking if you're making any of these 5 mistakes. A good quality schema will lead to a good quality output!
- [Build vs. Buy for Document Processing: How to Choose the Right Approach for Your AI Infra](https://reducto.ai/blog/build-vs-buy-ai-document-ingestion) (2025-04-15): If you're deciding whether to build your own document processing pipeline in house or buy with a vendor, this guide highlights some of the tradeoffs and why Reducto might be a good fit.
- [Introducing RolmOCR: A Faster, Lighter Open Source Document Model Built on olmOCR](https://reducto.ai/blog/introducing-rolmocr-open-source-ocr-model) (2025-04-03): We're excited to release RolmOCR, a drop-in alternative to olmOCR with a newer base model and other performance improvements.
- [Mistral OCR vs. Gemini Flash 2.0: Comparing VLM OCR Accuracy](https://reducto.ai/blog/lvm-ocr-accuracy-mistral-gemini) (2025-03-06): We benchmarked Mistral AI vs. Gemini Flash 2.0 to assess the accuracy of their OCR models after release. The results are intriguing as they don't appear to align with what was posted on Mistral's blog.
- [How We Started Working With Fortune 10 Enterprises](https://reducto.ai/blog/reducto-enterprise-sales) (2025-01-13): A behind the scenes look of how Reducto started powering ingestion for one of the world's largest companies.
- [State-of-the-art table parsing](https://reducto.ai/blog/sota-table-parsing) (2024-11-04): See how leading PDF parsers handle complex tables in RD-TableBench, an open benchmark of 1,000 hand-labeled examples covering merged cells, dense text, and irregular structures.
- [Announcing RD-TableBench: An Open-Source Table Benchmark](https://reducto.ai/blog/rd-tablebench) (2024-11-04): We released a new comprehensive open benchmark for table parsing.
- [Reducto raises $8.4 million to help LLMs read documents the way humans do](https://reducto.ai/blog/seed-round) (2024-10-02): Reducto now powers ingestion pipelines for some of the world's leading AI companies, and we're excited to share we've raised $8.4M in funding led by First Round to further our mission of making human data LLM-ready.
- [Introducing Reducto's Document API](https://reducto.ai/blog/document-api) (2024-02-27): We've spent the last few months building a powerful document ingestion for LLM workflows. We're excited to share more about what we've built.
- [The Real Cost of Manual Document Processing](https://reducto.ai/blog/the-real-cost-of-manual-document-processing) (2024-01-19): Explore the hidden costs of manual document processing, from time inefficiencies to error rates. Learn how Reducto's AI-driven solution can streamline these processes, saving businesses significant time and resources.

## Cookbooks

Templates and recipes for building with Reducto. This is an index — fetch each URL for the full recipe.

- [Turn loss runs and insurance submissions into underwriting-ready data](https://reducto.ai/cookbooks/reducto-loss-run-report-to-underwriting-data) (2026-08-06): Loss run reports are how carriers report claims history, and every one is laid out differently across dozens of pages of dense tables. This cookbook covers how you can create an end-to-end pipeline in Reducto that turns the whole report in one call into clean, structured tables you can load straight into code.
- [Split a law review article into sections and extract any part](https://reducto.ai/cookbooks/reducto-split-law-review-article) (2026-07-01): A law review article is long and highly structured: an abstract, a table of contents, and dozens of numbered sections, often across a hundred or more pages. Pulling one piece, say the abstract, first means finding where it lives before you can read it. With Reducto you parse the whole article once, split it into its constituent parts, and then run a targeted extract against only the pages that hold what you want. Three endpoints, one pipeline: Parse turns the PDF into clean structured text, Split maps the article into its sections, and Extract pulls the exact field you asked for with citations back to the source.
- [Parse any healthcare document into clean, structured text](https://reducto.ai/cookbooks/reducto-parse-healthcare-documents) (2026-07-01): Healthcare documents are some of the messiest inputs in any pipeline. A single patient's file might include a handwritten intake form, a scanned chart with checkboxes and body diagrams, a casualty card filled out under pressure, and a lab report full of dense tables, each laid out differently and often photographed or faxed. Retyping that by hand is slow, error prone, and does not scale. Reducto parses the entire document in one call and returns it as clean, structured text, handwriting transcribed and tables preserved, so your code or your model works with the data instead of the scan.
- [Pull every field from a handwritten property loss form](https://reducto.ai/cookbooks/reducto-extract-handwritten-property-loss-forms) (2026-06-29): First notice of loss forms come in handwritten, dozens of fields packed into a tight grid with corrections scribbled over the originals. Reducto reads the handwriting, skips the crossed-out edits, and returns every field as structured JSON with a citation on each value.
- [Turn a brokerage statement into structured account data](https://reducto.ai/cookbooks/reducto-extract-brokerage-statement-accounts) (2026-06-29): Brokerage statements bury account numbers, types, and balances in dense, multi-account tables that look different at every firm. Reducto parses those tables cleanly, then returns every account as structured JSON with a citation on each value.
- [Reconcile every transaction on a bank statement](https://reducto.ai/cookbooks/reducto-extract-bank-statement-reconciliation) (2026-06-19): Bank and brokerage statements arrive as PDFs and scans that spreadsheets can't read. Reducto's Extract turns one into a clean, typed list of every transaction, ready to load into a ledger, dashboard, or audit trail.
- [Extract structured data from patient intake forms](https://reducto.ai/cookbooks/reducto-extract-patient-intake-forms) (2026-06-19): Patient intake forms arrive as scans, faxes, and phone photos, every one laid out differently. Reducto's Extract reads them the way a nurse would, pulling demographics, insurance, and medication history into clean JSON with a source citation on every field.
- [Reducto + Databricks: a friendly parsing comparison](https://reducto.ai/cookbooks/reducto-databricks-comparison) (2026-06-15): Nine real cases where Reducto's parsing and Databricks ai_parse_document diverge, from spreadsheets and charts to signatures, tracked changes, and handwriting, and why those last-mile details decide whether downstream AI lands the right answer.
- [Pull every redline from a contract](https://reducto.ai/cookbooks/reducto-parse-redlined-legal-contracts) (2026-06-15): Redlined contracts are exactly the long-tail complexity that breaks template-based pipelines. Reducto reads every strikethrough, underline, and annotation as structured tags, so your team surfaces all 500+ revisions in code instead of page by page.

## Guides

Buyer guides, comparisons, benchmarks, and technical explainers for building with Reducto. This is an index — fetch each URL for the full guide.

- [Best Document Classification Software in 2026: APIs, Workflows, and Trade-Offs](https://reducto.ai/guides/best-document-classification-software-2026) (2026-07-30): Compare six document classification platforms across setup, confidence signals, workflow routing, human review, cloud fit, and production operations.
- [Best OCR APIs for Complex Documents in 2026](https://reducto.ai/guides/best-ocr-apis-complex-documents-2026) (2026-07-28): Compare six OCR APIs for complex documents across layout, tables, grounding, completion rates, cloud fit, and production workflows.
- [Document Workflow Automation: Architecture, APIs, and Production Patterns](https://reducto.ai/guides/document-workflow-automation) (2026-07-28): Learn how to build reliable document workflow automation using classification, parsing, splitting, extraction, editing, validation, and asynchronous delivery.
- [Enterprise IDP Evaluation Guide: How to Compare Document Processing Software in 2026](https://reducto.ai/guides/idp-enterprise-evaluation) (2026-02-06): Compare intelligent document processing software across accuracy, tables, citations, deployment, security, throughput, and cost with this enterprise IDP scorecard.
- [PDF to Text Conversion in 2026: Techniques, Tools, and Integration Guide](https://reducto.ai/guides/pdf-to-text) (2025-06-01): Compare techniques for converting digital and scanned PDFs into clean text, including embedded-text extraction, OCR, layout analysis, and production integration.
- [Intelligent Document Processing in 2026: Technology, Use-Cases, and Implementation](https://reducto.ai/guides/intelligent-document-processing) (2025-06-01): Learn how intelligent document processing combines classification, OCR, parsing, extraction, validation, and routing to automate document-heavy enterprise workflows.
- [Document Parsing: Turning Unstructured Files into Reliable, Structured Data](https://reducto.ai/guides/document-parsing-unstructured-files) (2025-06-01): Learn how document parsing converts PDFs, scans, spreadsheets, and images into structured text and JSON for search, analytics, automation, and LLM workflows.
- [Data Ingestion: Moving Unstructured Content into Your Analytics Stack](https://reducto.ai/guides/data-ingestion-unstructured-content) (2025-06-01): Build a document data ingestion pipeline that turns PDFs, scans, and spreadsheets into validated, structured outputs for analytics, automation, and RAG.
- [PDF Parser — What It Is, And Why You Need One](https://reducto.ai/guides/pdf-parser) (2025-06-01): Learn what a PDF parser does, why OCR alone struggles with complex layouts, and how to evaluate parsing software for structured data, RAG, and document automation.
- [Evaluating AWS Textract for PDF Parsing - Table Extraction](https://reducto.ai/guides/evaluating-textract-for-pdf-parsing) (2024-07-01): We evaluated AWS Textract's ability to parse documents.
- [Evaluating Azure Document Intelligence for complex PDFs - Table Extraction](https://reducto.ai/guides/evaluating-azure-document-intelligence-for-pdf-parsing) (2024-07-01): We evaluated Azure Document Intelligence's ability to parse complex PDF documents.
- [Evaluating LlamaParse Premium for PDF Parsing - Table Extraction](https://reducto.ai/guides/evaluating-llamaparse-for-pdf-parsing) (2024-07-01): We evaluated LlamaParse's ability to parse complex PDF documents.
- [Evaluating Unstructured.io for PDF Parsing - Table Extraction](https://reducto.ai/guides/evaluating-unstructured-for-pdf-parsing) (2024-07-01): See how Unstructured.io's Hi-Res mode scored 60.2% average table precision in RD-TableBench versus Reducto at 90.2%, including the errors behind the gap.
- [Evaluating Google Document AI for PDF Parsing - Table Extraction](https://reducto.ai/guides/evaluating-google-document-ai-for-pdf-parsing) (2024-07-01): See how Google Document AI scored 64.6% average table precision in RD-TableBench versus Reducto at 90.2%, including the structural errors behind the gap.

## More

- Engineering blog, indexed for LLMs: https://reducto.ai/engineering/llms.txt
- Sales / demos: https://reducto.ai/contact · Support: support@reducto.ai