Full-Time
Posted on 9/26/2025
Ingests data for LLMs and RAG
$90k - $140k/yr
San Francisco, CA, USA
In Person
| , , |
Reducto.ai helps large organizations handle big volumes of data by ingesting, parsing, and chunking documents so that information is easy to retrieve with large language models. Its system breaks down complex documents into meaningful chunks and extracts structured data, making it easier to feed relevant content into retrieval-augmented generation workflows that work with any vector database. The product works by processing pages, applying layout-based chunking and data extraction, and delivering organized content ready for LLM queries, with options for automatic feature parsing as an add-on. The company differentiates itself by offering enterprise-grade, scalable data processing with dedicated compute resources and tiered subscription plans based on page volumes, plus value-added features for large workloads. The goal is to help businesses improve RAG performance and decision-making by turning vast document collections into searchable, usable data.
Company Size
51-200
Company Stage
Series B
Total Funding
$108M
Headquarters
San Francisco, California
Founded
2023
Help us improve and share your feedback! Did you find this helpful?
Health Insurance
Dental Insurance
Vision Insurance
Unlimited Paid Time Off
Wellness Program
Parental Leave
Reducto raises $75M in Series B funding. Reducto, a San Francisco, CA-based AI document intelligence platform, raised $75M in Series B funding round. The round, which brought Reducto's total funding to date to $108M. was led by Andreessen Horowitz, with participation from existing investors Benchmark, First Round Capital, BoxGroup, and YCombinator. The company intends to use the funds to accelerate development across model research and product capabilities, and scale adoption across both enterprise and the next generation of AI teams. Led by Adit Abraham, co-founder and CEO, and Raunak Chowdhuri, co-founder and CTO, Reducto is a solution for turning complex documents into AI-ready inputs. Since its founding two years ago, the company has advanced a new standard for document understanding by combining traditional optical character recognition (OCR) with modern Vision-Language Models (VLMs), enabling systems to read documents as a human would. Customers range from AI-native startups, including Harvey, Rogo, and Scale AI, to global financial institutions and Fortune 10 enterprises. These companies use it to handle their most complex and mission-critical document workflow, such as, converting pdfs with redlines to text in legal workflows, extracting complex charts for financial due diligence, or high-stakes figure extraction for healthcare decisions.
Reducto, a startup integrating OCR with advanced AI to interpret documents, has secured new funding. This investment comes just six months after a previous round, highlighting the company's rapid growth and innovation in document data translation.
Reducto secures $75M led by Andreessen Horowitz. * Reducto secured $75M Series B led by Andreessen Horowitz, bringing total funding to $108M in under one year. * The AI document intelligence platform processes nearly one billion pages monthly for Harvey, Rogo, Scale AI, and Fortune 10 enterprises. * Andreessen Horowitz led the round with Benchmark, First Round Capital, BoxGroup, and YCombinator participating as existing investors.
Reducto, an AI document intelligence platform, raised a $75 million Series B round led by Andreessen Horowitz, bringing total funding to $108 million. The company, founded two years ago, combines OCR with Vision-Language Models to enhance document understanding. Reducto's platform processes nearly a billion pages monthly for clients like Scale AI and Fortune 10 enterprises. The new funding will accelerate model research and product development, expanding adoption across enterprises and AI teams.
Reducto has also launched two key improvements - a new agentic OCR framework, which automatically reviews Reducto's outputs, catching mistakes and making corrections through a multi-pass VLM framework, similar to having a human in the loop, and smart cost savings for simpler pages.