Ocrolus Raises $US24M to scan financial documents
Ocrolus, a New York startup that utilises machine learning to capture data from financial documents, has announced $US24M in new funding will be used to automate underwriting workflows for lenders and banks, and expand into new verticals.
The company claims to offer the only automation platform that analyzes financial documents with over 99% accuracy. Ocrulus says it has quintupled in size since April 2018 and now counts hundreds of financial services companies among its customer base.
Ocrolus offers a turnkey solution hosted on AWS. The platform reviews bank e-statements, scans, and cell phone pictures of printed statements and generates custom-tailored analytics, ultimately geared toward tracking account balances, detecting money movement, and identifying missing or fraudulent information. Its API plugs into Salesforce and other customer relationship management platforms, and it guarantees a baseline level of security with 256-bit AES encryption, SSL authentication, and multifactor authentication.
“Historically, image recognition software has not been accurate enough to automate financial review work completely. Machines struggle to parse semi-structured documents like bank statements and pay stubs, and lower quality images like cell phone pictures,” the company said in a statement.
“Ocrolus solves these inherent problems and eliminates data entry and cleansing tasks, using its human-in-the-loop validation engine to analyze every file with over 99% accuracy. Turnkey by design, Ocrolus ingests images of any format or quality, and returns actionable data directly into its customer's back-office systems in minutes. The platform powers business processes for hundreds of customers in the financial services sector.”
"Sometimes humans are better than robots," said Sam Bobley, Co-founder and CEO of Ocrolus. "We combine machine processes with live human intelligence to provide customers with a complete solution. The capital will be used to develop workflows for new document types, and sharpen our fraud detection and analytical capabilities."