Conor Burke spent most of his career in the back office of a large bank in Ireland. His team was tasked with digitizing board processes that were costing the bank millions of dollars every year and making them fraud-free. The biggest challenge, he says, is figuring out how to eliminate the human element without reducing risk and fraud control.
Inspired by this, Burke and his twin brother Ronan Burke launched Inscript, an AI-powered document fraud detection service. Built for fraud, risk and operations teams in the fintech and financial industries, inscribe pipelines trained on hundreds of millions of data points to return results, Ronan says.
“Tedulous document reviews add friction to the account opening and underwriting process, but automation alone is not the solution,” Ronan told TechCrunch in an email interview. And it’s a whole package that helps them approve more customers faster.”
Record analytics, categorize, and correlate data with financial onboarding documents, using AI-powered fraud detection to differentiate between received and received documents. Document details such as names, addresses and bank statement transactions are automatically digitized to generate customer risk profiles that include bank statements and snapshots of transactions.
Last September, Inscript rolled out a credit analytics and bank statement automation unit that collects most of the data points needed to make credit decisions, including cash flow details from bank statements, transaction analysis and account balance analysis. Ronan says he can run an Inscript and then retrieve key details in seconds, including names, addresses, dates, transactions and salaries.
Image Credits: write
In its features, Inscript is similar to other anti-fraud tools such as Resistant AI (which raised $16.6 million in October 2021) and Smile ID (which raised $7 million in July of the same year). Ronan argues that it stands out in its AI-first approach, but relies on original data gathered through previous partnerships with customers.
“We’ve seen fraud in our space and document automation companies try to build a perfect solution right out of the gate without talking to customers — but then they’re shut down. They can’t get past the cold start problem; they can’t build a product from the ground up because they can’t access the data their customers are using,” Ronan said. Here comes the first rule of machine learning: start with data, not machine learning. If you don’t have a good dataset, you’re wasting your time. By choosing the wrong model or training a model on data, it may not perform as you expected.
AI is not perfect by any stretch of the imagination – history shows that much is true. For example, during the pandemic, fraud detection systems embedded in unusual behavior were confused by new transaction and spending habits. Elsewhere, automated algorithms designed to detect welfare fraud are prone to error and designed in ways that penalize the poor.
But putting aside the veracity of Ronan’s claims, there’s clearly something about Inscript’s platform that attracts high-profile clients. TripActions, Ramp, Bluevine and Shift are among the startup’s clients.
And investors won. Just this week, Inscript closed a $25 million Series B funding round led by Threshold Ventures, with participation from Crosslink Capital, Foundry, Uncork Capital, Box founder Dillon Smith and Intercom founder Des Traynor. The infusion brings the startup’s total revenue to $38 million, after a $10.5 million Series A round closed in April 2021.
Perhaps it’s the comparative simplicity with which the Inscript solution can work. As Ronan rightly notes, Inscript solves the problem of having to build a fraud detection solution in-house or hire a large data science team.
“AI and machine learning models use as much data as possible, but each individual company is limited to their own data set. So an in-house solution cannot be as effective as pulling from multiple data sources,” said Ronan. “Companies are instead collaborating with document fraud detection solutions: Criminals commit fraud in a variety of ways, and those solutions are pulling data from their customers to quickly identify coordinated attacks and emerging trends.”
Fear can also help. A recent study found that the average US fintech loses $51 million to fraud, statistician Ronan quoted me in our interview.
“An increasingly digital, geographically fragmented and fast-paced world makes it harder than ever to know who to work with – leaving companies unsure of which customers they can trust,” said Ronan. “Fintech has been built for the online world, but traditional financial institutions are challenged to break out of legacy systems and embrace true digital transformation. And they must do it all while reducing fraud and friction to achieve competitive customer experiences.”
Asked about expansion plans, Ronan says Inscript may double its 50-person workforce in the next 12 to 18 months.
