Retab Raises $US3.5M for Document AI Platform
San Francisco-based startup Retab has secured $US3.5 million in pre-seed funding to develop its document processing platform that extracts structured data from PDFs and scanned documents using artificial intelligence.
The funding round was backed by VentureFriends, Kima Ventures, and K5 Global, alongside notable investors including former Google CEO Eric Schmidt, Datadog CEO Olivier Pomel, and Dataiku CEO Florian Douetteau.
Retab's platform allows developers to define data schemas while the system handles document processing through automated model selection and prompt engineering. The company claims its approach can reduce costs compared to alternative solutions, though specific performance metrics could not be independently verified.
The startup was founded by engineers Louis de Benoist (CEO), Sacha Ichbiah, and Victor Plaisance, who previously built document automation tools for logistics workflows. The 10-employee company emerged from stealth mode with the funding announcement.
Retab positions itself as a developer-focused platform that works with existing large language models from OpenAI, Google, and Anthropic rather than developing proprietary models. The platform features what the company describes as "self-optimising schemas" and multi-model consensus mechanisms for accuracy validation.
The company claims its system can achieve "up to 100x" cost reductions compared to competitors and cites unnamed customers in trucking and financial services achieving high accuracy rates. These performance claims could not be independently confirmed.
The document AI market includes established players like Amazon's Textract, Google's Document AI, and Microsoft's Form Recognizer. Retab enters a competitive landscape where enterprises seek to automate manual document processing while ensuring accuracy and compliance.
Dataiku CEO Florian Douetteau, an investor in the round, stated that converting document-based operations into structured data is essential for AI adoption across the economy, particularly for quality control and cost efficiency.
Retab plans to expand its platform to process web content and develop integrations with additional automation platforms. The company envisions serving as middleware between unstructured data and AI agents requiring structured information.
The startup aims to become foundational infrastructure for what it terms "vertical AI" applications across industries processing document-heavy workflows.