AI verifies mortgages ‘within seconds’ in Perpetual, Lakeba and Microsoft partnership

AI is being deployed across the financial services industry at a rapid pace, with Perpetual saying it can now verify mortgage documentation in about 40 seconds.

Sydney tech firm Lakeba is automating document verification across some of Australia’s biggest financial institutions, slashing processing times from about 12 minutes to within seconds, and will soon be deployed across multi-billion dollar transactions.

Lakeba’s DoxAI has developed and deployed its AI tools onto Microsoft’s Cloud to accelerate digitisation at ANZ, Macquarie Bank and now Perpetual.

Perpetual Corporate Trust chief executive Richard McCarthy said the 137-year-old firm was using AI to run quality checks on securitised loans and mortgages and planned to later extend its application to payments, which involves billions of dollars of transactions a day.

“There’s still a large paper based business and process in the banking and financial services industry of which we are an intermediary for and as a business, we’re always focused through what we call our ‘three E’s to C strategy’ – how do we make our business, our clients and the market more effective, more efficient, more economical, while improving and ensuring that we focus on the ever increasing risk of cybersecurity,” Mr McCarthy said.

“We have a technology which enables our clients to repeatable and scalable solutions that ensure our loan security documents are complete, accurate and ultimately enforceable. And we are really starting to create immense amounts of process improvement, accuracy and time benefit.”

Artificial intelligence is being deployed across the financial services industry at a rapid pace.

Mr McCarthy said Perpetual planned to extend the technology’s use to payments in the second quarter. “That’s been a long time coming because … we’ve got seven legacy systems being replaced by a cloud SaaS (software as a service) solution, which processes hundreds of billions of dollars of payments – thousands of transactions a day. So it has to be 100 per cent accurate”.

Financial services have been one of the earliest and biggest adopters of AI. And while Australian Securities & Investments Commission chair Joe Longo said there was “plenty of scope for making the best use of our regulatory toolkit” to regulate AI, it did not mean more needed to be done to rein in the technology.

“Even as AI ‘leaps forward,’ at a rate never seen before, questions around transparency and explainability become paramount if we’re to protect consumers from harm – intended or not,” Mr Long said.

“It isn’t fanciful to imagine that credit providers using AI systems to identify ‘better’ credit risks could (potentially) unfairly discriminate against those vulnerable consumers. And with ‘opaque’ AI systems, the mechanisms by which that discrimination occurs could be difficult to detect. Even if the current laws are sufficient to punish bad action, their ability to prevent the harm might not be.”

But Mr McCarthy said Perpetual’s use of AI was strictly about verifying documentation, a process which typically takes a human around 12 minutes to do. The AI has slashed this process to about 48 seconds, Mr McCarthy said.

“Our process is really around process improvements rather than credit quality, of which the banks are basically providing that loan to an individual, determining that loan is fit for purpose and that loan can be serviced by them within their credit parameters.

“We’re one step down in terms of actually really trying to ensure once that loan is taken out, by the bank off or non bank financial institution, that it is 100 per cent accurate.

“What happens sometimes, you know, you have a land title registry, the certificate of title, the mortgage document, there’ll be a spelling error in one of those documents, and therefore the name is wrong, and really the entire loan contract needs to be accurate, verified and then enforceable. That’s where these improvements are coming from.”

Lakeba chief executive Giuseppe Porcelli also said AI’s use in document verification meant it wasn’t making decisions in which any bias could be an issue. Rather, he said, it was ensuring documentation was more correct, achieving about 97 per cent accuracy.

“I will use an Italian expression – it depends on how many drinks or how good the lifestyle of the human had the day before? Because you know, humans sometimes are subjective to something else in their life. The person can be distracted for any reason,” Mr Porcelli said.

“So the accuracy, in my words, can be really competitive when it comes to human versus the ability of a machine to discover issues in a digital or physical document.

“We firstly create products by making sure that we deeply understand what the challenges are, especially when it comes to workflows and process compliance.”