Document processing automation startup Mindee today emerged from stealth with $14 million in series A funding led by GGV Capital, with participation from Alven, Serena Capital, and Bpifrance through its Digital Venture fund. The company plans to use the funding to expand its presence in the U.S. and Europe and further develop its machine learning and platform technologies.

Despite the growing move toward digital transformation, HR professionals, accountants, government employees, and clerks take hours to manually enter information into their systems. Surveys show that employees spend somewhere around 150 hours a year searching for files in cabinets and 30% to 40% of their time looking for lost or misplaced paper. The average cost of each misfiled document is $125, and some organizations lose a document roughly every 12 seconds.

San Francisco, California-based Mindee, which was founded in 2018, leverages AI and APIs to capture data from scans of expense management, accounting, insurance, employee onboarding, loan applications, and underwriting documents. Mindee imports the data into first- and third-party systems with algorithms trained on a library of use cases and workflows.

“All businesses, regardless of size, deal with multiple paper-heavy processes. Think about back-office processes, expense management, accounts payable automation, accounting processes, loan applications, insurance policy registration — the list goes on,” CEO Jonathan Grandperrin told VentureBeat via email. “Mindee’s technology makes the entire process faster and easier for all parties involved — the employee or applicant submitting and the company personnel who review, approve, and reimburse.”

AI-powered processing

Mindee’s AI algorithms first analyze the document image from a visual perspective to locate where pertinent data might be hidden and then read only that subpart of the image. That’s as opposed to optical character recognition technology, which reads all the words in a document and captures data based on language processing.

“Mindee combines both computer vision and a traditional optical character recognition-based natural language processing pipeline to parse documents. The advantage of this is very high precision — less false positives than traditional machine learning — and very strong geography robustness,” Grandperrin said. “[We’ve] tapped into large global datasets from different types of clients to eliminate potential business or industry biases … More than 60 countries [are] represented in our datasets.”

Mindee competes with services from Microsoft, Amazon, Adobe, and others in the multibillion-dollar intelligent document processing market. For example, Google offers Document AI (DocAI) platform, a console for AI-driven document processing hosted in Google Cloud. Startups in the space include Anvil, Flatfile, and Rossum, among others, some of which focus on particular industry verticals.

But the company, which has an office in France, says it has approached potential customers in the finance, research, insurance, government, health care, and logistics industries. Mindee’s document-parsing API is currently available with a free-of-charge trial.

The growing adoption of automation technologies in the enterprise is likely to benefit Mindee in the months to come, according to Grandperrin. A 2020 McKinsey survey found that 31% of businesses have automated at least one function and 66% were piloting solutions to automate at least business process, up from 57% two years earlier.

“Since January 2021, Mindee has doubled the number of employees, with a plan to end 2021 with 30 employees. Employees are now located in France, Canada, and the U.S. In 2022, the plan is to increase to 45 employees internationally,” Grandperrin said. “There are more than 50 software companies deploying Mindee’s API in production, processing millions of pages monthly for their clients. The estimate of monthly active end-users — those using customer software — is approximately half a million.”