Introduction
Ocrolus builds AI software that reads and analyzes the financial documents lenders rely on to make credit decisions — bank statements, pay stubs, tax forms, and similar paperwork — turning scanned or photographed pages into structured, verified data. Co-founder and CEO Sam Bobley started the company in New York in 2014, initially building the product himself before growing it into a platform used across small-business, mortgage, and consumer lending.
Ocrolus is a useful complement to Brel’s AI in finance coverage of credit-decisioning companies, since it addresses a step that comes before any underwriting model can run: getting reliable, structured data out of documents that arrive in wildly inconsistent formats and quality.
What the company does
Ocrolus’s platform classifies incoming financial documents, extracts and validates the data on them, analyzes cash flow and income, and flags likely fraud, using a combination of optical character recognition, machine learning, and what the company calls Human-in-the-Loop review — humans checking and correcting model output — to push accuracy toward the near-100% level lenders require before making a credit decision. In 2023, Ocrolus introduced Instant, a fully automated, machine-only processing option for cases that no longer need the human-review step.
Who it serves
Ocrolus sells to lenders and fintechs rather than to individual borrowers. Its own 2023 and 2024 press releases name customers including Brex, Enova, LendingClub, PayPal, Plaid, SoFi, CrossCountry Mortgage, Better Mortgage, Deephaven Mortgage, and PennyMac Financial Services, spanning small-business funding, mortgage underwriting, and consumer lending. Ocrolus’s own company page states it analyzes roughly 750,000 credit applications a month across more than 3,700 active user accounts, with use cases the company says have expanded over time to include tax preparation and Medicaid eligibility document review alongside its core lending customers.
Company background
Ocrolus was founded in 2014 in New York by Sam Bobley, who has served as the company’s co-founder and CEO since its start; other sources report additional early collaborators, but Ocrolus’s own communications consistently credit Bobley as co-founder and CEO without naming further founders. Bobley has said he began building the company’s first product in his parents’ kitchen at age 22. Ocrolus raised an $80 million Series C round in September 2021, led by Fin VC, at a valuation the company said was north of $500 million, taking its total disclosed funding above $100 million; a further Series C-II round closed in February 2023.
Product and AI capabilities
Ocrolus’s AI stack is built to handle financial documents regardless of format or scan quality, supporting more than a thousand document types, according to its own materials, and pairing machine-learning extraction with human reviewers who check and correct the model’s output before data is passed on to a lender’s own underwriting system. This Human-in-the-Loop design — rather than a fully automated pipeline from the outset — reflects the company’s stated goal of near-100% accuracy on inputs as varied as a clean statement from a major bank and a blurry photo of a paper statement from a small community lender. The 2023 launch of Instant added a fully machine-only processing tier for customers willing to trade the human-review step for faster turnaround on higher-confidence documents.
Key developments
Ocrolus was founded in 2014 and grew its annual recurring revenue from about $1 million in 2018 to roughly $20 million by mid-2021, according to CEO Sam Bobley, while adding customers such as Brex, PayPal, Plaid, and SoFi. It raised an $80 million Series C in September 2021 at a valuation above $500 million, and a Series C-II round in February 2023. In September 2023, it launched Instant, a fully automated document-processing product, and the company has since repositioned its platform under a “Vertical AI” framing, reporting more than 3,700 active user accounts as of its most recent company materials.
Why it matters
Ocrolus is a useful corrective to AI-in-lending narratives that focus only on the underwriting model itself: a credit decision is only as good as the data feeding it, and a large share of that data still arrives as scanned or photographed paper documents of wildly inconsistent quality. Ocrolus’s decision to keep a human-in-the-loop step, rather than pursuing full automation from day one, is also a useful illustration of how document-AI vendors serving regulated lending have approached the trade-off between speed and verified accuracy, offering a fully automated option like Instant only once its underlying models had enough training volume to support it with confidence.
Sector context
Ocrolus is part of Brel’s AI in finance coverage as a document-data infrastructure vendor, positioned upstream of underwriting-model companies such as Upstart and Zest AI and adjacent to account-connectivity providers such as Plaid, one of its own customers. Readers evaluating any AI-driven lending product should generally understand both where its account data and its document data come from, and Ocrolus is a common answer to the latter.
Sources and references
This profile draws on Ocrolus’s own company page and press releases, and independent reporting on its funding history.
- Ocrolus — “About Ocrolus | Our Vertical AI Journey” company page (ocrolus.com)
- Ocrolus — “Ocrolus launches new, cost-effective Instant solution…” (2023)
- TechCrunch — “Ocrolus lands $80M at a $500M+ valuation…” (2021)