Fundraising Fox

Extend

YC W23

New York City, US · Founded 2023 · 25 employees · 10 known investors

Extend builds infrastructure that processes unstructured data at scale for applications in healthcare, real estate, insurance, and finance. The platform helps companies make messy, inconsistent documents usable and accessible across critical industries.

Also known as CrowdView · Extend AI · extend.ai

Founders & leadership· Y Combinator alumni (W23)

Extend was founded in 2023 by Kushal Byatnal and Eli Badgio.

KBKushal Byatnal
Kushal Byatnalin𝕏Co-founder · CEOKushal Byatnal was an early engineer at Brex and later cofounded Stir, a fintech platform for creators. He is now building internal tools infrastructure at Extend.
EBEli Badgio
Eli Badgioin𝕏Co-FounderCo-founder of Extend, a platform that processes unstructured data at scale for applications in healthcare, real estate, insurance, and finance.

Investors · 10

Also in the syndicate · 1

March Capital

Funding

SEC filings, press & company announcements

Source: company announcements and press reports — follow each round's link for the claim.

Company profile

researched Aug 2026

Extend provides document processing infrastructure for AI agents and data pipelines. Its APIs parse, extract, and split documents, converting PDFs and other unstructured files into structured, high-quality data that downstream applications and agents can consume. The company states its systems process millions of pages every day.

Beyond core models, Extend ships a toolkit intended to make document pipelines production-ready: confidence scoring backed by a multi-pass review agent that flags uncertain outputs, selectable processing modes trading off latency, cost, and maximum accuracy, a Composer optimization agent that refines extraction schemas from uploaded examples, multi-step document workflows with parsing, splitting, extraction, validation, and routing plus versioning and durability, and a Studio interface for schema iteration, evaluations, and regression testing by domain experts. Developer access is available via Python and TypeScript SDKs and a CLI, distributed through the extend-ai package.

The company publishes benchmark results comparing its parsing against alternatives. On RealDoc-Bench, which it describes as testing production documents where layout, reading order, and field relationships affect answer quality, Extend Parse 2.0 reports 95.7% field-level answer accuracy across 1,359 prompts and 581 documents, ahead of Extend Parse 1.0 (90.4%), Gemini 3.5 Flash (89.04%), Azure DI (88.8%), and AWS Textract (70.5%).

Business model

Extend sells document processing APIs and tooling to engineering and AI teams, offering a free trial entry point alongside demo-led enterprise sales. The product can be consumed as a hosted cloud service or deployed self-hosted on a customer's own infrastructure, which supports customers in regulated industries that must keep sensitive documents in-house.

Traction

Extend states that it processes millions of pages daily. Brex reports Extend powers key document workflows across 30,000 customers; Flatiron Health reports replicating six months of work in two weeks and scaling across a network covering 5 million people with cancer; Vendr reports Extend produced the best results in a competitive evaluation.

Latest developments

Extend announced Parse 2.0, described as a layout-first parsing API, and published accompanying benchmark results on document Q&A, layout, cost, latency, and vertical-specific performance.

Full profile — market position, technology, go-to-market, geography, history

Market position

Extend competes in document parsing and extraction against cloud services such as Azure Document Intelligence and AWS Textract, open-source tooling, and general-purpose foundation models. Customer testimonials from Brex, Vendr, and Flatiron Health describe Extend outperforming alternatives in evaluations, and Extend's own benchmark places Parse 2.0 first on its RealDoc-Bench document Q&A test.

Extend positions its accuracy on difficult production documents as its primary differentiator, publishing head-to-head benchmark comparisons against its prior version, foundation models, and cloud document AI services. It also emphasizes surrounding production tooling — confidence scoring, evaluation and schema-iteration interfaces, workflow orchestration, and an optimization agent — rather than models alone, plus self-hosted deployment and compliance certifications for regulated buyers.

Technology

The platform combines parsing, extraction, splitting, and classification APIs with orchestration and evaluation tooling. Features include a layout-first parsing engine (Parse 2.0), a multi-pass review agent for confidence scoring, multiple processing modes for latency-, cost-, or accuracy-optimized runs, the Composer agent for automated schema refinement, durable versioned multi-step workflows, and a Studio environment for evals and regression detection. Security and compliance include SOC 2, HIPAA, and GDPR, regular third-party penetration testing, and self-hosted deployment.

Go-to-market

The company markets through a self-serve free trial and developer documentation with Python, TypeScript, and CLI clients, combined with a book-a-demo motion for enterprise buyers. Customer stories from Brex, Vendr, and Flatiron Health and published benchmark results serve as primary proof points, and prospects are described as running competitive bake-offs against other vendors, open-source tools, and foundation models.

AI and engineering teams building document-driven applications, including companies in finance, real estate, logistics, and healthcare. Named users include Brex, Vendr, and Flatiron Health, and the company says it is trusted by Fortune 500 organizations in regulated industries.

Geography

Extend is headquartered in New York City and serves customers including Fortune 500 organizations, with cloud and self-hosted deployment options.

History

Founded in 2023 and based in New York City, Extend has developed successive generations of its parsing technology, with Parse 2.0 succeeding Parse 1.0 as its layout-first parsing API. The company reports processing millions of pages per day and counts AI and technology teams including Brex, Vendr, and Flatiron Health among its customers.

Compiled by commissioned research from 1 cited public sources — announcements, filings, and press listed under research sources below.

Key figures

latest reported
Brex customers served by Extend-powered document workflowsJan 202530,000 customers
HeadcountAug 202632
Long Array overall mean per-document accuracyJan 202599.2%
PoliTax Split F1 lift over raw modelJan 202528.4 points
RealDoc-Bench Document Q&A accuracy (Extend Parse 2.0)Jan 202595.7%

Company-reported or press-reported figures, each dated to when it was claimed — not independently audited.

Competitors · 9

by search overlap
Docsumo92 shared keywordsDocsumo provides APIs and document processing platform to extract data from unstructured documents across industries including insurance, banking, and logistics. The platform enables businesses to automate document data capture with 90%+ touchless processing capabilities.
Koncile59 shared keywordsKoncile is a French SaaS platform that automates data extraction from unstructured business documents using OCR technology combined with large language models. The solution serves enterprises across accounting, logistics, and document management by converting scanned documents, PDFs, and images into structured, actionable data.
NanoNets59 shared keywordsNanonets provides AI agents that automate document-heavy, complex business processes by extracting structured data from documents like invoices and purchase orders, then executing workflows across finance, operations, and healthcare systems. The platform integrates with existing enterprise tools and uses a proprietary OCR model to handle edge cases that undermine agent reliability.
LlamaIndex57 shared keywordsLlamaIndex provides document parsing and AI agent workflows that automate document-based administrative operations such as invoice matching, contract data extraction, form processing, and compliance reporting. It serves enterprise operations and developer teams building retrieval-augmented generation (RAG) and agent-based applications.
Blue Prism56 shared keywordsSS&C Blue Prism provides an enterprise agentic automation platform (WorkHQ) that unifies human workers, digital workers and AI agents, combining robotic process automation with AI decision-making and built-in governance, security and orchestration. It targets large, complex and highly regulated organizations, with products including AI agent templates, an AI governance gateway and the Chorus orchestration tool.
Netezza50 shared keywordsIBM is a global technology company whose business spans enterprise software (including Red Hat, HashiCorp, and Confluent), IT infrastructure such as mainframes, servers, and storage, and IT consulting services. The company is also investing heavily in quantum computing and AI-based enterprise offerings, including its Lightwell open-source software security clearinghouse and the Anderon quantum wafer foundry.
V748 shared keywordsV7 is an AI platform that automates document-heavy workflows for private markets, insurance, and real estate investment teams, covering tasks such as investment diligence, reporting, submission ingestion, and underwriting. It is used by investment and financial services firms to process complex deal documents at scale.
Evernote43 shared keywordsEvernote is a note-taking and productivity platform that helps individuals and teams capture, organize, and access information across devices. The service includes AI-powered features for transcription and search, task management, and collaborative editing for personal and professional use.
Lido40 shared keywordsLido is a document data extraction tool that pulls information from documents, including handwritten and scanned files in any layout or language, validates it against connected systems, and automates downstream workflows.

Companies competing with Extend for the same Google search keywords, organic and paid, via search-intersection analysis.

Timeline · 1

launches, deals, and filings
Jan 2025
Launch of Parse 2.0 layout-first parsing API

Extend released Parse 2.0, a layout-first document parsing API, announced on its website and blog.

source ↗

Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.

In the news

Research sources · 1

primary sources listed

1 public sources were cited for this profile; the first-party ones are listed here.

Frequently asked questions

What does Extend do?
Extend provides document parsing, extraction, and splitting APIs that turn unstructured documents into data for AI agents.
Who founded Extend?
Extend was founded by Kushal Byatnal, Eli Badgio in 2023.
Who are Extend's investors?
Extend's investors include Hawk Hill Ventures, Homebrew, Y Combinator, B Capital, Character, Commerce Ventures, FinTech Collective, Innovation Endeavors and 1 more.
Where is Extend headquartered?
Extend is headquartered in New York City, US.