# Extend

> AI-powered document processing platform providing APIs and tooling for parsing, extraction, classification, splitting, and editing with 99%+ accuracy.

## About

Extend transforms unstructured documents into structured, high-quality data. The platform provides production-ready APIs that handle the hardest document processing use cases — from dense healthcare forms to 2,000+ page loan packets — so teams can ship reliable pipelines in minutes, not months.

Extend uses a hybrid computer vision + vision-language model pipeline that routes each document element to purpose-built models, delivering unmatched accuracy across 25+ file types and 100+ languages.

## Core APIs

- **Parse API**: Convert unstructured documents into LLM-ready markdown. Advanced layout detection for tables, checkboxes, images, handwriting, and signatures. Multiple performance modes for speed, cost, or accuracy.
- **Extract API**: Extract structured data from documents into any schema. Scales to 1,000+ page files with smart chunking & merging. Precise bounding box citations on every extracted value. Low-latency fast mode for real-time use cases.
- **Classify API**: Classify documents into pre-defined categories. Cost-optimized classification at scale. Multimodal memory system that learns from past examples.
- **Split API**: Segment multi-document files into individual subdocuments. High-precision splitting on 2,000+ page files. Instance detection and intelligent boundary handling.
- **Edit API**: Detect form fields and fill them programmatically. Supports checkboxes, signatures, text fields, tables, dropdowns, multi-line paragraphs, and character-per-box inputs. Dynamic natural language filling or deterministic template-based filling.

## Platform Features

- **Confidence Scoring**: Multi-pass review agent checks every output, flags potential errors with confidence scores on every extracted value.
- **Composer Agent**: Optimization agent that automatically refines schemas and improves accuracy — skip manual prompt trial-and-error.
- **Document Workflows**: End-to-end orchestration for complex pipelines with versioning and durability.
- **Studio & Evals**: Iterate on schemas, run evals, catch regressions from one intuitive interface. Empowers domain experts beyond CLI scripts.
- **Agentic OCR**: Advanced OCR powered by vision-language models trained for real-world challenges like handwriting and specialized elements.
- **Fast Mode**: Toggle between processing modes optimized for speed, cost, or accuracy.

## Industry Focus

- **Healthcare**: CMS-1500s, EOBs, prior authorizations, medical records, patient charts, physician handwriting
- **Financial Services**: Expense processing, loan origination, KYC, claims processing, invoices, receipts, articles of incorporation
- **Real Estate**: Leases, PSAs, deeds, tax statements, seller disclosures, closing documents, signature detection
- **Supply Chain & Logistics**: Bills of lading, proof of delivery, freight invoices, fuel statements, packing lists

## Benchmarks and source data

- **Benchmark hub**: https://www.extend.ai/benchmarks
- **RealDoc-Bench report**: https://www.extend.ai/resources/realdocbench
- **LongArray-Extract report**: https://www.extend.ai/resources/long-array-extraction-benchmark
- **PoliTax Split report**: https://www.extend.ai/resources/document-splitting-benchmark
- **Parse 2.0 and RealDoc-Bench launch**: https://www.extend.ai/resources/parse-2-and-realdocbench-launch
- **RealDoc-Bench source repository**: https://github.com/extend-hq/realdoc-bench
- **RealDoc-Bench document Q&A dataset**: https://huggingface.co/datasets/Extend-AI/RealDoc-Bench
- **RealDoc-Bench layout dataset**: https://huggingface.co/datasets/Extend-AI/RealDoc-Bench-Layout/
- **LongArray-Extract dataset**: https://huggingface.co/datasets/Extend-AI/LongArray-Extract
- **PoliTax Split dataset**: https://huggingface.co/datasets/Extend-AI/PoliTax-Split

## Security & Compliance

- SOC 2 certified with regular third-party penetration testing
- HIPAA compliant infrastructure
- GDPR compliant
- Self-hosted deployment option — run on your own infrastructure
- Trusted by F500 companies in regulated industries

## Pricing

- **Pay As You Go**: Free — 10,000 credits included, full product access, chat support
- **Scale**: $500/month — 50,000 credits/month, volume discounts, Slack support, custom data retention, BAA add-on
- **Enterprise**: Custom pricing — self-hosted deployments, custom MSA/DPA/SLAs, SSO & SAML, advanced RBAC, custom model fine-tuning, dedicated support

## Customers

Trusted by teams at Brex, Vendr, Flatiron Health, Opendoor, Column Tax, Checkr, HomeLight, Nudge Security, AbstractOps, Mercury, Square, Zillow, Chime, Tesorio, Nuvocargo, and more. Brex alone serves 30,000+ customers powered by Extend.

## Funding

$17 million in seed and Series A funding led by Innovation Endeavors, with participation from Y Combinator, Homebrew, Character, and angel investors including Scott Belsky and Guillermo Rauch.

## Links

- Website: https://www.extend.ai
- Documentation: https://docs.extend.ai
- Dashboard: https://dashboard.extend.ai
- Demo: https://dashboard.extend.ai/demo
- OCR Arena: https://www.ocrarena.ai/battle
- Blog: https://www.extend.ai/resources
- Pricing: https://www.extend.ai/pricing
- Twitter: https://x.com/ExtendHQ
- LinkedIn: https://www.linkedin.com/company/extend-app/