Reducto raises $24.5M Series A to help enterprises unlock unstructured data

Real-world impact from AI means working with messy, real-world data. We witnessed this problem firsthand, and last year we set out to build the most accurate parsing pipeline in the industry—leveraging the best of traditional computer vision and new Vision-Language Models (VLMs)—to help companies turn their most complex documents into precise, LLM-ready inputs with state-of-the-art accuracy.

That's why Reducto has quickly become the go-to solution for many of the world's best AI teams. Since our launch, thousands of companies, including Airtable, Scale AI, and FAANG enterprises, have trusted Reducto to parse hundreds of millions of documents.

We're now focused on making Reducto the definitive platform for leveraging unstructured data end-to-end. Building on our industry-leading parsing capabilities, we've extended support to power comprehensive workflows—including document splitting, intelligent classification, precise structured extraction, and more. Our upcoming platform will integrate all of these capabilities to help any enterprise build accurate pipelines with their unstructured data.

Today, we're thrilled to announce that we've raised a $24.5M Series A led by Benchmark, bringing our total funding to $33 million. This marks an exciting milestone as we accelerate our mission of making human data LLM-ready, and unlock the next phase of our growth.

---

Parsing without compromises – our new Agentic OCR framework

From the start, we designed our parsing pipeline to be accurate, reliable, and built to scale. We’ve relentlessly improved it since to tackle the challenging long tail of document scenarios—expanding support for new file types, handling complex structures like equations, and ensuring consistent real-world reliability through precise bounding boxes.

Today, we're excited to release two key improvements that make our parsing pipeline even better.

1. Agentic OCR Framework
   We’re unveiling a new Agentic OCR framework—a step change in document processing. This agentic approach automatically reviews Reducto’s outputs, catching mistakes and making corrections through a multi-pass VLM framework, similar to having a human-in-the-loop. We intend to continue extending this framework to help our customers unlock near perfect parsing accuracy with their most challenging documents.

2. Smart Cost Savings for Simpler Pages
   We've updated our pipeline to automatically discount simpler pages that can be parsed accurately without sacrificing fidelity. With zero loss in accuracy, Reducto now halves the cost of processing simpler pages, enabling our customers to always benefit from best-in-class accuracy without maintaining separate pipelines for different complexities.

---

Production-grade document pipelines for any use case

Parsing is just the first step. Companies are already leveraging Reducto's API endpoints to build end-to-end pipelines for intelligent splitting and structured extraction.

We will soon be unveiling a new platform that takes this capability even further by enabling complex workflows that integrate all of Reducto's models—parsing, splitting, classification, and extraction—into a unified, easy-to-manage solution.

With more and more companies looking for ways to use AI to accelerate their work, our team is looking forward to launching a user-friendly interface that anyone can utilize to automate their data processing and pipelining.

We’ve been fortunate to partner with some of the largest enterprise companies across industries—finance, healthcare, tech, and legal—parsing over 250M+ pages of documents to help solve critical bottlenecks for their AI teams.

Companies like Vanta and Airtable trust us to safely and accurately extract insights from massive volumes of documents—powering smarter automation and enhancing their products.

---

Unlocking the next phase of our growth

We’re fortunate to work with a group of incredible investors across our seed and series A. This round was led by Benchmark, alongside existing investors First Round Capital, BoxGroup, and Y Combinator, and brings our total funding to $33M.

We take our role as the ingestion team for our customers seriously, and this round will help us serve more teams with an ever-improving product. Reducto started with vision models for documents, expanded to offer state-of-the-art parsing across all enterprise file types, and will soon empower anyone to create and build fully functioning processing pipelines across all use cases.

We have a lot more in store, and would love to meet you if you’re excited about joining us on our mission. You can view our open roles here or learn more at reducto.ai.

More from us soon.
– Adit and Raunak