Parsewise Launches API to Extract Structured Data Across Multiple Unstructured Documents

Parsewise, a Y Combinator P25 startup founded by Greg and Max, has launched an API that converts large collections of unstructured files — such as PDFs, spreadsheets, and transcripts — into schema-compliant structured data. The platform traces every extracted value back to word-level citations across source documents, addressing a key validation challenge that existing tools leave unsolved. Unlike retrieval-augmented generation approaches, Parsewise performs exhaustive search across all relevant documents rather than sampling, and uses layered models for parsing, search, and decision-making. The founders draw on backgrounds at Palantir and Bain, and say the system is model- and cloud-agnostic, with support for private network deployment. Target users include tech teams handling complex data ETL pipelines in fields such as insurance and finance.