CredoLab closes in on Series-A raise
The Singapore-based A.I. company is extending its mobile-based credit scores to new markets.
CredoLab expects to raise at least $3 million this quarter in a Series A round, says CEO and co-founder Peter Barcak. He says the company should have a term sheet finalized this month.
The company, founded in 2016, has already raised $3.1 million in seed equity and convertible notes, but the Series A round will put a valuation on the company—although Barcak declined to comment on what that might be.
Forum and GoBear are among the company’s publicly known earlier backers.
The newly raised capital will go first to geographic expansion, and secondly to research and development of CredoLab’s technology.
The company develops digital scorecards for banks, insurers, and other lenders such as consumer and auto finance companies from smartphone device metadata.
The service is designed for lending to people who lack access to physical bank branches or who are from countries with no credit bureau or otherwise low penetration rates of financial services.
CredoLab’s artificial intelligence-based algorithm crunches over 1 million features from opt-in smartphone metadata to find the most predictive behavioral patterns before converting them into credit scores.
We measure the willingness of someone to payPeter Barcak, CredoLab
Barcak declined to describe those patterns – this is the company’s intellectual property. In general, a phone’s metadata includes data about call times, duration, and numbers dialed. But with borrower consent, it can also mean scouring the device for all of the apps and other services on a phone.
Metadata does not include information itself but details about the information. For example, such a scan might aggregate information about when messages are read or sent, but do not read the actual messages.
By sticking to metadata, CredoLab avoids touching personal identity information. Nonetheless, by combining a vast number of metadata points, the software can draw conclusions, in close to real time. In CredoLab’s case, the output does not measure the ability of someone to repay a loan. “We measure the willingness of someone to pay,” Barcak said.
Although Barcak says CredoLab is careful to avoid touching personal or sensitive information, its algorithms and the huge number of data points deliver decisions that are not easily explainable.
For this reason, the company hasn’t yet marketed its services in the U.S. There, regulators require a bank or lender must be able to give a reason to someone rejected for a loan or other product. FICO scores and credit bureau information gives U.S. banks the data they need to tell someone why they didn’t qualify for a loan.
The money now being raised will go in part to developing technology to help the company be able to identify more precisely why it’s A.I. might recommend denying a user’s application.
The bulk of the money will go to helping CredoLab develop new markets, particularly in Africa and Latin America; deepen its business in Southeast Asia and India; and expand its clientele into non-financial corporations, such as e-commerce companies and other “big database owners”, as Barcek calls them.
“We signed our first ‘analytics as a service’ client in January in Indonesia,” he said. “We help businesses monetize their data, by being able to better understand their customers’ needs and ultimately build lasting relationships with them.”