This Hong Kong startup develops AI to help banks find dirt on SMEs

Think finding accurate and free information on the internet is easy? Think again.

Photo credit: Tech in Asia

What was once dubbed the information superhighway has deteriorated into a nightmarish traffic jam, filled with repetitive stories, misinformation, and a language barrier that runs right down the middle.

That’s what Megan Chau, co-founder and CEO of Hong Kong-based risk assessment startup inRiskable, discovered when her company started developing natural language processing solutions for banks in 2021.

Despite banks having access to services like the Bloomberg Terminal and Factiva, she found that they were still going to Google and Baidu – and printing out the results on hard copy – to conduct due diligence.

“We realized a lot of banks have trouble in finding a customer’s background, especially for SMEs,” Chau tells Tech in Asia. “There is so much news online, and the information is flowing so rapidly. They just can’t keep track of it.”

InRiskable co-founders Megan Chau and Kenyon Wong. / Photo Credit: InRiskable

Banks lack the personnel to track daily news online, and because of that, many don’t have enough data to understand what’s happening with their clients and customers.

What banks typically do is watch media sites for news while they are onboarding a company and scan again about once a year, says inRiskable co-founder and CTO Kenyon Wong.

Chau and Wong saw an opportunity to fill the gap by scanning the open web for credible news about companies. News about smaller firms, which aren’t consistently covered by major media outlets, is especially in demand.

Focus on real-time risk reporting

In 2022, the company pivoted to focus on what banks really need: real-time data and analysis that lets them know if their clients present a risk.

“They want a tool that can help them to construct the report and also help them to read through all the data in one second,” says Wong. They want a report three minutes later.

InRiskable uses its own fine-tuned, large language model to analyze the data from a white list of over 730 websites, which includes media outlets, regional courts, and press releases. While the firm targets freely available data, if the information is behind a paywall and the client requests it, then inRiskable will pay to access it.

Competitive pricing with big-name rivals

Possible applications in crypto trading