Berbix raises $9M to grow next-generation biometric identity verification based on signals aggregation

Berbix raises $9M to grow next-generation biometric identity verification based on signals aggregation

Startup Berbix has announced a $9 million funding round for its new approach to cloud-based biometric selfie and ID document checking service, with plans to invest the funds in team and customer growth.

Prior to founding the company, co-founders Steve Kirkham and Eric Levine led Airbnb’s Trust and Safety team. Levine told Biometric Update in an interview with the two founders that during that time they became “intimately familiar with the challenges of the existing identity verification market. We used pretty much every solution under the sun, and found that it left a lot to be desired.”

They formed Berbix in 2018 to provide a new generation of ID verification solutions that they say are faster, more effective and secure. They are focussing on customers using the technology for age verification, driver or rider verification, KYC checks, and fraud mitigation, such as fighting chargebacks and account takeovers. They also say it can reduce overhead costs.

The Series A round was led by Mayfield, with existing investors Initialized Capital, Y Combinator and Fika Ventures participating, according to the announcement. The company says its growth has continued to exceed expectations during the COVID-19 pandemic, and it is now processing the same number of transactions each day that it was in a month a year ago.

Kirkham and Levine found the identity verification providers they were familiar with from Airbnb were slow, expensive, and cumbersome, sometimes taking three to five minutes to return results after a human review. While Levine acknowledges those times have decreased, the instantaneous results that work within a “synchronous flow” require instant results.

The ability to deliver those instant results, the customer experience that creates, plus accessibility to smaller companies are what makes Berbix a new generation offering, the founders say.

Levine says they’re fraud-focussed approach includes thinking about the type of fraud that used to defeat Airbnb’s defences, and Kirkham explains that by considering how fraudsters get around fraud prevention measures, they were able to make real improvements in detecting fake IDs, to the point of deterministic decisions (rather than a risk score). Berbix is concentrating on the U.S. and Canada to start with, and uses what Levine says is a novel approach to catch “factors more” fake IDs than competitors in company testing.

“Even though we have industry standard support for IDs across the world today, we have intentionally decided to go ridiculously deep across the United States and Canada,” Levine says.

The company does this by aggregating signals, and using a combination of machine learning and artificial intelligence technologies along with what Kirkham calls “a series of painstakingly curated, hand-written rules.”

“The combination of both allows us to predict with incredibly high accuracy, and especially with those hand-written assertions, deterministically be able to reject fake IDs without any human intervention, and still within our fully instantaneous response,” he says.

Other companies in the space, including newer entrants, all mimic in-person ID checks, according to Kirkham. They go through the steps of ascertaining the validity of the document, subject comparison, and data validation.

“We do that too, but that’s not our primary approach,” Kirkham explains. That’s our secondary, corroborative approach. Our primary approach is leveraging the machine-readable components on the ID, in addition to the human readable fields on those same IDs. And due to the machine-readable components on most of the IDs across the United States and Canada, we’re able to detect fraud, or fake IDs, or IDs that are impossible to have been legitimately issued by a DMV, with this incredibly high, and in most cases deterministic accuracy.”

This “image and data-based approach” returns instantaneous results that are conclusive more than 95 percent of the time,

Berbix also works with clients to determine the next best steps, and provides tools for manual review.

Context and data are important, and Berbix will never have them to the extent its customers do, Levine says. “So instead we built best in class tools to empower our customers to make the decisions they need on those inconclusive results that flow back to both their systems and their system in real time.”

For facial biometrics, Berbix uses an algorithm enhanced with “home-grown ML and AI-based solutions to improve upon what used to be just off-the-shelf,” Kirkham says. The company targets unique technologies to detect very specific fraud vectors where fraudsters try to circumvent biometrics, with methods such as checking the image background.

Kirkham describes Berbix as “a privacy-first company that is actually sitting at the identity and fraud,” and cites its compliance with BIPA as evidence of that respect for privacy.

The company cites pro-COVID forecasts from McKinsey that the market for ID verification-as-a-service will grow from $10 billion in 2017 to between $16 billion and $20 billion in 2022, so growth opportunities should be plentiful if Berbix can convince the market that its technology works as advertised.

Related Posts

Article Topics

 |   |   |   |   |   |   |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Most Read This Week

Featured Company

Biometrics Research

Biometrics White Papers

Biometrics Events

Explaining Biometrics