Oscilar emerges from stealth to fight transaction fraud with AI

Image credit: William_Potter/Getty Images

Confluent co-founder Neha Narkhede today announced a new fintech company, Oscilar, which is developing an “AI-powered” platform to help financial institutions protect online transactions from fraud and theft.

Oscilar is entirely self-funded, backed by $20 million that Narkhede and the company’s other co-founder, Sachin Kulkarni, contributed themselves. Narkhede says they chose not to take external funding so they could “quickly build and scale the company” when it goes public.

“Sachin and I were in companies with some of the world’s best engineering talent, but we were both baffled by how cumbersome and, frankly, inefficient the prevailing risk decision technology was,” Narkhede told TechCrunch in an email interview. “We firmly believe that fraud and risk decision technology needs an overhaul, and we’ve been working under the radar for the past two years to build a company to solve this problem.”

Before co-founding Oscilar, Kulkarni was a senior director of engineering at Facebook, where he led the division that helped build Facebook’s internal private cloud. Meanwhile, in addition to launching Confluent, Narkhede was a LinkedIn executive overseeing the network’s efforts to scale its data pipeline.

With Oscilar, Narkhede and Kulkarni sought to address what they characterize as a “dramatic increase” in the risk of online transaction fraud for companies. The pandemic further accelerated the explosion in online transactions over the past decade, they note, leading to a corresponding increase in fraud, credit and insurance risks.

To their point, more people are using digital payment technology than ever before, with a 2022 World Bank survey finding that two-thirds of adults worldwide now make or receive a digital payment. At the same time, the risk has actually increased. Juniper Research expects online payment fraud losses to eclipse $200 billion by 2025.

Of course, Oscilar is far from the first to attempt to tackle the enormous problem of payment risk. Nivelo recently launched with $2.5 million in seed funding for its platform that protects ACH transactions. Sardine has raised $51.5 million to develop anomaly detection technology for fintech payments. Other players in the space include Fraugster, Bouncer (which Stripe acquired in 2021) and Hawk AI, the last of which marketed its products specifically towards banks.

But Narkhede claims that many of the platforms out there don’t cut it.

“Most companies’ investments in fraud and risk protection have not kept pace with the proliferation of online transactions,” Narkhede said. “In a nutshell, existing technologies rely on incomplete and outdated information about user behavior.”

What makes Oscilar different, says Narkhede, is the platform’s heavy reliance on AI and machine learning. But not just any AI – Narkhede claims that Oscilar’s AI, developed in-house, requires much less first- and third-party data about past fraud incidents from customers to train machine learning models.

“We integrate aggregated anonymous risk signals from the entire network of customers into our machine learning models, [and] our machine learning models are continuously trained and tested,” explained Narkhede. “This process is fast as we have automated the feedback loop that updates our models. As a result, they automatically get smarter over time.”

Oscilar also provides a toolset that risk operations teams can use to create new risk models, test them, monitor and deploy the models without having to write code. This frees these teams from being dependent on engineers for intervention and technical support, says Narkhede.

“With Oscilar, our customers have visibility into the model and determine all parameters and data that go into the model,” she continued. “In fact, as part of our onboarding process, we take customer data and feed it into the model to make sure it’s working properly and we’re not introducing biases. Oscilar keeps people in the loop throughout the process and shares the results with the company to ensure that the performance is good and that the model works as it should.”

Given the competition in fraud defense, Oscilar definitely has its work cut out for it. Funding could also prove a challenge – especially given the recent high-profile bank failures. According to CrunchBase, Venture investments in fintech companies in 2022 reached $81 billion as of December 2022, which was down 41% from the peak in 2021 at $137 billion.

But in what could potentially be a positive sign, Narkhede claims Oscilar is already working with “dozens” of fintech clients.

“We’ve quietly brought in dozens of engineers with decades of experience from Facebook, Google, Uber and Confluent who have deep expertise in building highly scalable AI-powered computing systems,” she said. “The public launch of Oscilar allows us to engage with an even wider range of potential customers and partners.”