Startups > Rossum

Rossum

Rossum automates document data extraction for enterprises using proprietary AI.
Series A $104M total Founded 2017 United Kingdom
Rossum is an AI-powered intelligent document processing (IDP) platform that automates data extraction from transactional documents like invoices, purchase orders, and bills of lading. Built on Rossum Aurora, a proprietary transactional large language model trained on millions of business documents, the platform helps enterprises eliminate 90% of manual document processing work while maintaining data security through in-house AI models. It serves over 450 global customers across finance, logistics, and manufacturing, including Bosch, Siemens, and PepsiCo, with human-in-the-loop validation that continuously improves accuracy without relying on third-party LLMs.
Problem solved
Finance teams spend excessive time manually extracting and validating data from business documents, with each document taking hours to process despite varying layouts and formats.
Target customer
Enterprise finance departments, shared service centers, and large organizations processing high volumes of transactional documents (invoices, POs, bills of lading) across multiple languages and formats.
Website LinkedIn Crunchbase
Founders
T
Tomáš Gogar
CEO & Co-Founder
AI PhD student with Bachelor's in software engineering and Master's in AI from Czech Technical University; named to Forbes 30 under 30.
P
Petr Baudiš
CTO & Chief AI Architect
Computer Science researcher with background in theoretical CS and AI; co-founder of Ailao; core Git developer and creator of open-source AI for Go, cited in Google's AlphaGo paper.
T
Tomáš Tunys
CSO & Co-Founder
Chief Strategy Officer; AI PhD student at founding.
Funding history
Seed Undisclosed February 2017 Led by Miton
Seed $8.5M November 2019 Led by LocalGlobe · Seedcamp, Elad Gil
Series A $100M October 2021 Led by General Catalyst · LocalGlobe, Seedcamp, Miton, Elad Gil
Total raised: $104M
Industries
Artificial Intelligence Document Management Machine Learning SaaS Software
Pricing
Subscription-based SaaS model starting at $18,000/year ($1,500/month) for Starter plan; Business and Enterprise tiers are custom-priced based on document volume and workflow complexity. Flexible subscription terms range from 12-36 months with additional costs for ERP integrations and add-ons.
Notable customers
Bosch, Siemens, Panasonic, Flexport, Morton Salt, The Master Trust Bank of Japan, Cushman & Wakefield, PepsiCo, Wolt
Integrations
ERP systems (custom integrations available); specific integration partners not extensively detailed in public materials.
Website
rossum.ai/
Competitors
Nanonets
General-purpose document AI platform with lower cost positioning suitable for SMBs, but less specialized for transactional documents than Rossum's domain-specific Aurora model.
Amazon Textract
Cloud-based document extraction service using AWS's general LLM, lacks Rossum's specialized transactional model and human-in-the-loop continuous learning.
Infrrd
Competitive IDP platform but relies on template-based approaches and third-party LLMs rather than Rossum's proprietary template-free transactional model.
Docsumo
Lower-cost alternative focused on SMBs and startups, with simpler setup but less enterprise-grade features and accuracy than Rossum's Aurora model.
Why this matters: Rossum represents a compelling approach to document AI by building proprietary models trained exclusively on business documents rather than relying on general-purpose LLMs—addressing both accuracy and data security concerns for enterprise customers. With $104M in funding, 450+ enterprise clients, and a 2021 Series A led by General Catalyst, Rossum has established itself as a credible challenger in the enterprise IDP space, though its high cost and long onboarding cycles limit addressable market compared to more accessible alternatives.
Best for: Large enterprises and shared service centers processing high volumes of invoices, purchase orders, and financial documents across multiple languages who can justify the $18K+ annual commitment and 3-month onboarding period.
Use cases
Invoice Processing at Scale
Finance departments processing thousands of invoices monthly in varying formats and currencies. Rossum Aurora automatically extracts line items, amounts, tax breakdowns, and vendor details with 90% reduction in manual work. Master Trust Bank of Japan reduced document validation time from 1.5 hours to 50 seconds per document.
Multi-Language Document Handling
Global enterprises receiving purchase orders and bills of lading in 276+ languages. Rossum's proprietary model handles language variations and handwriting without data leakage to external LLM providers, critical for companies with sensitive supply chain data.
Continuous Accuracy Improvement
Organizations requiring document processing accuracy that improves over time. Rossum's human-in-the-loop validation system learns from corrections, continuously refining the AI model without requiring template updates or retraining.
Alternatives
Amazon Textract Choose if you need low-cost, AWS-native document extraction without specialized financial document handling or if you're comfortable with general-purpose LLM approaches.
Nanonets Choose if you're a small-to-mid-size business needing faster, simpler setup and lower cost, though with less enterprise support and specialized transactional accuracy.
Docparser Choose if you need a lightweight, affordable document parsing tool for basic extraction needs rather than enterprise-grade AI-powered processing with continuous learning.
FAQ
What does Rossum do? +
Rossum is an AI-powered intelligent document processing platform that automatically extracts and validates data from transactional documents like invoices, purchase orders, and bills of lading. It uses a proprietary transactional language model called Aurora trained on millions of business documents to achieve high accuracy while eliminating 90% of manual work. The platform supports 276 languages and includes human-in-the-loop validation that continuously improves accuracy without data leakage to third-party AI services.
How much does Rossum cost? +
Rossum's Starter plan begins at $18,000/year ($1,500/month). Business and Enterprise tiers are custom-priced based on the volume of documents processed annually and workflow complexity. Additional costs apply for ERP integrations and add-ons. Subscription terms range from 12 to 36 months.
What are alternatives to Rossum? +
Top alternatives include Amazon Textract (AWS-native, lower cost but less specialized), Nanonets (faster onboarding for SMBs), Docparser (lightweight extraction), Infrrd (template-based IDP), and Docsumo (SMB-focused). Choice depends on document volume, budget, required accuracy, and preferred implementation timeline.
Who uses Rossum? +
Rossum serves over 450 organizations globally, including large enterprises like Bosch, Siemens, Panasonic, PepsiCo, Wolt, and The Master Trust Bank of Japan. It's best suited for enterprise finance departments, shared service centers, and organizations processing high volumes of transactional documents across multiple languages.
How does Rossum compare to Amazon Textract? +
Rossum's proprietary Aurora model is specifically trained on business documents and excels at financial document understanding (multi-currency, tax handling, line item matching), while Textract uses general-purpose AWS LLMs. Rossum maintains data security through in-house models and includes human-in-the-loop learning, whereas Textract operates as a simpler extraction service. Rossum costs more but delivers higher accuracy for transactional documents and continuous improvement over time.
How long does Rossum take to implement? +
Enterprise onboarding typically takes 3 months for shared service centers and large organizations. Rossum offers 'Premier' and 'Signature' onboarding plans as part of their implementation process. Setup is part of their enterprise DNA, with professional services engagement as standard.
What makes Rossum different from competitors? +
Rossum built its own transactional language model (Aurora) specifically for business documents rather than relying on general-purpose or third-party LLMs. This proprietary approach eliminates data leak risks, improves accuracy on financial documents, and enables continuous learning through human-in-the-loop validation. It's a Gartner Magic Quadrant Challenger, providing enterprise credibility.
Tags
document processing intelligent document processing IDP invoice automation AI machine learning enterprise automation RPA LLM transactional data extraction