Unstructured Technologies, Inc.
Rocklin, US · Founded 2022 · Delaware corporation · 9 known investors
Open-source and API-based platform that converts unstructured documents into AI-ready structured data for LLM and RAG pipelines.
Also known as Unstructured · Unstructured-IO · Unstructured.io
Founders & leadership
Unstructured Technologies, Inc. was founded in 2022 by Brian Raymond.
Board
Investors · 9
Also in the syndicate · 6
Reported raises · per SEC filings
Form D private placements$68.1M disclosed across 2 rounds · 2023–2024
▶$43.2MraisedMar 2024 · 20 investors · Other TechnologyRule 506(b)
- Karan MehandruDirector
- Enrique SalemDirector
- Brian RaymondExecutive Officer, Director
- Tim TullyDirector
- Offering amount
- $43.2M
- Amount sold
- $43.2M
- First sale
- Feb 2024
- Incorporated
- Corporation, Delaware, 2022
- Federal exemptions
- 06b
▶$24.9MraisedJun 2023 · 19 investors · Other TechnologyRule 506(b)
- Enrique SalemDirector
- Brian RaymondExecutive Officer, Director
- Karan MehandruDirector
- Offering amount
- $26.5M
- Amount sold
- $24.9M
- First sale
- Jun 2023
- Incorporated
- Corporation, Delaware, 2022
- Federal exemptions
- 06b
Source: SEC EDGAR Form D. Amounts as filed; amended filings shown once at their latest values.
Company profile
researched Aug 2026Unstructured Technologies, Inc. develops tooling that ingests, preprocesses and partitions unstructured documents into structured, machine-readable formats for AI and machine learning workloads. Its central asset is the open-source Python library "Unstructured," hosted on GitHub under the Unstructured-IO organization and distributed via PyPI under the Apache 2.0 license. The library automatically detects file types and converts raw inputs such as PDFs, HTML, Word documents, PPTX, images and plain text into structured outputs including JSON, text chunks, tables and associated metadata that preserve semantic and structural information. Sources report support for 54 file types.
Alongside the open-source components, the company offers a hosted service and API that scale the same processing for enterprises, plus more than 15 production-grade data connectors that link to enterprise data sources and destinations and keep pipelines continuously updated. The stated purpose is to remove the manual data-engineering work required before natural language data can be used with large language models, and to support retrieval-augmented generation pipelines by supplying source-grounded, verifiable data intended to reduce model hallucinations.
The company also serves U.S. government defense and intelligence customers. It is listed as a member of the SOSSEC consortium with capabilities in data discovery, visualization and analytics; data mining; intelligence production; and intelligence analysis, exploitation and dissemination, and holds UEI CUSYKN4GZ8C7 and CAGE code 9BHF4.
Founding story
Unstructured Technologies was founded in July 2022 by Brian Raymond, Matt Robinson and Crag Wolfe, who had previously worked together at Primer AI. According to the sources, they started the company to address the absence of efficient tooling for connecting, transforming and staging unstructured data for machine learning, having observed data scientists spending excessive time building custom preprocessing pipelines. The initial open-source project focused on producing clean training and evaluation data for NLP, with early cleaning functions and staging code compatible with Hugging Face models.
Business model
Unstructured combines an open-source core with commercial offerings. The Unstructured Python library is distributed free on GitHub and PyPI under the Apache 2.0 license, while the company sells access to a hosted platform and API, including a single API for using natural language data with LLMs and a set of production-grade data connectors that maintain continuously updated pipelines. The company also derives revenue from U.S. government work, having been awarded Phase I and Phase II Small Business Innovation Research contracts from the U.S. Air Force and U.S. Space Force, and a reported $2M AFWERX TACFI award. Its privacy policy describes the service as intended for business/commercial use, with account, payment and content data collected from customers.
Commercial revenue is generated from a hosted platform and API tools sold under enterprise agreements or terms of service, with payment card and banking data collected for service payments. Government revenue comes from SBIR Phase I and Phase II contracts awarded by the U.S. Air Force and U.S. Space Force and a reported $2M AFWERX TACFI award. The core Python library itself is open source under Apache 2.0.
Traction
As of July 2023, the library had over 700,000 downloads and was integrated into more than 2,400 GitHub repositories. The company reported key partnerships and a growing customer base, technology co-developed with the open-source community, commercial enterprises and select U.S. Government defense and intelligence organizations, and Phase I and Phase II SBIR contracts from the U.S. Air Force and U.S. Space Force. Later signals include a $2M AFWERX TACFI award and an expanded Microsoft Azure integration.
Latest developments
Press mentions listed on a secondary-market profile point to continued activity through 2026: a $2M AFWERX TACFI award announced February 18, 2026 for multimodal data pipelines and test-and-evaluation frameworks for generative AI; Teradata announcements in March and May 2026 relating to autonomous processing of text, images and audio at enterprise scale; and an expanded Microsoft Azure integration announced June 3-4, 2026. The company's privacy policy was last updated July 8, 2026, indicating the service remains in operation. Shares in the company are listed for secondary trading on EquityZen under the ticker symbol UNST.
▸Full profile — market position, technology, go-to-market, geography, history, risks & controversies
Market position
Described in the July 2023 funding announcement as a leading provider of LLM data preprocessing solutions, evidenced by download and repository-integration counts. Its investor and angel base spans AI infrastructure companies (LangChain, Weaviate) and data platforms (MongoDB Ventures), and it is grouped by one secondary marketplace with data and AI software companies such as DataRobot, H2O.ai, Harness, ThoughtSpot and Dataminr, though the sources do not name direct competitors.
Sources emphasize an open-source-first approach developed with the community, with the library integrated into more than 2,400 GitHub repositories and over 700,000 downloads as of July 2023, plus broad file-type coverage (54 types) and an extensible API for customizing partitioning strategies. A second differentiator described is the company's defense and intelligence credentials, including SBIR Phase I and Phase II contracts with the U.S. Air Force and U.S. Space Force and an advisory board drawn from the Joint AI Center, the Defense Innovation Unit, and In-Q-Tel/AWS National Security.
Technology
An open-source Python library that ingests documents, automatically detects file types, and partitions content into structured elements such as text chunks, tables and metadata, emitting formats like JSON for downstream AI pipelines. It supports 54 file types (PDF, HTML, Word, PPTX, images, text), is licensed Apache 2.0 on GitHub and PyPI, and exposes an API allowing users to customize partitioning strategies while remaining lightweight enough for local or cloud deployment. Commercial layers add a single hosted API and more than 15 production-grade connectors to enterprise data sources and destinations, with the stated aim of supporting retrieval-augmented generation and reducing LLM hallucinations by supplying source-grounded data.
Go-to-market
Distribution begins with the free open-source library on GitHub and PyPI, which drives developer adoption and integration into downstream projects, followed by conversion to the commercial API and hosted platform. The company also pursues government business through SBIR/AFWERX contracting and consortium membership (SOSSEC), and builds partnerships and integrations with adjacent infrastructure vendors such as Microsoft Azure, with investor-affiliated ecosystem players including MongoDB Ventures and angels from LangChain and Weaviate.
Enterprises and data science/ML engineering teams building generative AI, RAG and NLP applications on internal document stores, developers using the open-source library, and U.S. Government defense and intelligence organizations.
Geography
U.S.-based. Sources give differing locations: Rocklin, California and San Francisco, California (encyclopedia entry), Sacramento, California (funding announcement), and a listed address of 7580 Horseshoe Bar Road, Loomis, CA 95650 (SOSSEC consortium record). The company states it drives adoption of LLM-native data preprocessing worldwide.
History
The company was founded in July 2022 by Brian Raymond, Matt Robinson and Crag Wolfe, former colleagues at Primer AI. The open-source library's first pre-release appeared on September 6, 2022, with the first stable version (0.2.0) on September 26, 2022, shortly before the public release of ChatGPT, which accelerated adoption. The v0.5 series began on February 28, 2023, adding a Wikipedia connector for the ingest CLI among other changes. In July 2023 the company announced $25 million in combined Seed and Series A financing, released a single API and more than 15 data connectors, and added defense-oriented advisors. Later press mentions indicate continued activity: a $2M AFWERX TACFI award (February 2026), work referenced alongside Teradata's platform announcements (March and May 2026), and an expanded Microsoft Azure integration (June 2026). Sources conflict on headquarters (Rocklin, San Francisco, and Sacramento, California are variously cited) and on the Series A lead investor (Madrona per the funding announcement; Thrive Capital per one encyclopedia entry).
Risks & controversies
Public sources are inconsistent on basic facts, which limits verification: headquarters is variously given as Rocklin, San Francisco and Sacramento, California, with a Loomis, California address on a government-consortium listing, and the July 2023 Series A lead is reported as Madrona in the funding announcement but as Thrive Capital in an AI-generated encyclopedia entry. An SEC EDGAR entity page returned no filings data. The business also depends on an open-source project whose commercial conversion is not documented in the sources, and on U.S. defense contracting, which is subject to award cycles.
Compiled by commissioned research from 6 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
Timeline · 7
launches, deals, and filingsPress mention of Unstructured expanding its integration with Microsoft Azure to support enterprise AI workflows.
Press mention: Unstructured awarded a $2M AFWERX TACFI to advance multimodal data pipelines and test & evaluation frameworks for generative AI.
$2M source ↗
Unstructured announced $25 million across Seed and Series A rounds; Series A led by Madrona and seed led by Bain Capital Ventures. Karan Mehandru (Madrona) and Enrique Salem (Bain Capital Ventures) joined the board.
$25M source ↗
Retired General Michael Groen (former Director, Joint Artificial Intelligence Center), Mike Brown (former Director, Defense Innovation Unit), and Ryan Lewis (In-Q-Tel and AWS National Security veteran) joined the advisory board.
Product update introducing a single API for using natural language data with LLMs, plus more than 15 production-grade data connectors for building continuously updated data pipelines.
Start of the v0.5 series, adding a Wikipedia connector for the ingest CLI along with other enhancements and fixes.
First stable version of the open-source library released on September 26, 2022, following a 0.0.1.dev0 pre-release on September 6, 2022; introduced basic partitioning of unstructured documents into structured formats.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
Legal entities · 1
corporate structure▸Research sources · 6
primary sources listed
- Unstructured Closes $25 Million in Funding | The SaaS Newsthesaasnews.com · web
6 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Unstructured Technologies, Inc. do?
- Open-source and API-based platform that converts unstructured documents into AI-ready structured data for LLM and RAG pipelines.
- Who founded Unstructured Technologies, Inc.?
- Unstructured Technologies, Inc. was founded by Brian Raymond in 2022.
- Who are Unstructured Technologies, Inc.'s investors?
- Unstructured Technologies, Inc.'s investors include Bain Capital Ventures, Madrona Venture Group, Menlo Ventures.
- How much funding has Unstructured Technologies, Inc. raised?
- Unstructured Technologies, Inc. has disclosed $68.1M raised across 2 rounds.
- Where is Unstructured Technologies, Inc. headquartered?
- Unstructured Technologies, Inc. is headquartered in Rocklin, US.