Startup pioneers AI to detect accounting fraud

Transparently.AI identifies red flags far in advance of a corporate collapse.

Startup Transparently.AI has developed a pioneering technology that detects accounting manipulation and fraud using artificial intelligence (AI), predicting the risk of corporate collapse almost instantly.

This can help a bank make informed decisions about a loan application, an asset manager to evaluate their portfolio risk, or an investor to prevent financial losses by shunning a business that’s about to fold.

“Imagine if you could predict the collapse of Enron or Wirecard three years before it happened,” Mauro Sauco, chief technology officer (CTO) and co-founder of Transparently.AI, told Singapore Business Review. “Imagine having models that could tell you that — that’s what we do.”

Transparently.AI’s Manipulation Risk Analyzer performs what a forensic accounting team normally takes at least a month to do — analyse the financial risk of a company for any year and generate highly accurate reports down to a few seconds.

The AI’s machine learning algorithm combines the power of advanced data science techniques and big data to unearth accounting red flags that the human eye might miss.

Regulators can also use the tool to monitor corporate governance and enforce remedial actions or rescue plans to strengthen their regulatory environment and avoid reputational damage.

“Our goal is to detect these issues much earlier, more efficiently, and reduce potential losses from accounting manipulation,” said Sauco, who worked as a technical director for the Office of the CTO at Google for six years until April 2023.

Transparently.AI gives subscribers access to more than 80,000 listed stocks domiciled and traded in public markets globally. For smaller enterprises, the platform features a self-service portal that grants access to a curated network of data sets from one or two countries.

“We have built a huge array of models that are designed to replicate the behaviour of forensic accountants, short sellers, credit analysts, equity analysts and more,” Transparently.AI CEO and co-founder Hamish Macalister said.

These models - almost 200 of them - are then combined in an AI system [that] can access the financial data of nearly 100,000 listed companies globally, he added. The Singapore-based startup has built a database of “known manipulators” — companies that have been involved in some form of accounting fraud.

“Our system can then learn from those the combinations of features that signal a company is likely to be engaged in accounting manipulation — and in its extreme form, outright fraud,” said Macalister, who is a fintech inventor and data scientist.

“These combinations can be highly complex, may deviate from traditional approaches that market participants take, and are constantly evolving,” he added.

Fraud detection is an “ideal problem” for AI, which can model the real world more accurately than non-AI methods and adapt as companies use new forms of manipulation, according to Macalister.

“We're not only finding the things that any forensic accountant can find, but we also discover things they don't even know exist yet,” Sauco added.

To generate the MRA’s actionable insights, users simply enter queries for specific companies into the platform. The system generates a risk score and rating to identify high-risk companies along with their issues and red flags. Users have the option to download a detailed 20-50 page report for each company, which is instantly available with the click of a button.

Future-proofing the business
WorldCom, Enron, Wirecard, subprime, and the like are just a tiny, though popular, sample of financial scandals rooted in accounting manipulation. In Singapore, Noble Group, a commodity trader, was fined $12.6m for misleading financial statements from 2012 to 2016.

They represent the tip of the proverbial iceberg with respect to the financial damage caused by such fraud. Transparently.AI estimates trillions of dollars in direct financial losses due to accounting fraud by listed companies alone.

Four of five companies doctor financial figures each year and 10% commit outright securities fraud, according to a whitepaper posted on the startup’s website.

There are no platforms similar to Transparently.AI’s technology. The startup wants to become a formidable player against future rivals “by having very large enterprise customers who are happy with our services and by maintaining technology that is ahead of its time,” Sauco said.

The company recently got $3m in pre-series A funding from investors such as Franklin Templeton and Move Asset Management, allowing it to expand its team and manage its growing client portfolio.

The startup already had a “healthy portfolio” of customers even before it installed a sales team at the start of the year, Sauco said. “It always has been just founders and relationships. And those seem very organic.”

WIth a proper sales team and a new chief revenue officer, the company expects to further expand its customer base.

Transparently.AI is investing more in technology by hiring in-house developers to future-proof its service. It has also hired data scientists to work on a “revolutionary new technology” that the startup is developing with Google — one that it says will define their next-generation legacy. He declined to elaborate.

“We are very well aware that AI technology is developing at an incredible pace,” Sauco said. “We have built a unique solution to a very painful problem. We have an extensive development pipeline that will help deepen and widen our competitive moat.”