Moula relies on Microsoft AI to power SME lending

Fintech disruptor Moula has transitioned to a Microsoft Azure based platform and is using data, AI and machine learning to assess loan applications from small businesses.

Instead of a lengthy application process and weeks-long waiting period, Moula’s approach means a 10-minute online application, with API access to business data which determines credit worthiness, can result in more funds being made available within 24-48 hours.

Microsoft Azure platform as a service and a range of Azure cognitive services now provide Moula’s digital foundations. Moula worked with Microsoft partner, BizData to establish its Azure based data warehouse using Azure SQL DB, implemented Power BI to run business analytics, and leverage Azure AI and machine learning capabilities to assess data and analyse risk.

According to Justin Green, Chief Product and Strategy Officer Moula; “If you don’t have the benefit of assets, like the family home to put up as collateral, then your ability to borrow money from the bank is pretty limited.” Moula instead allows a small business to liberate the value in its own financial data by using that information to assess the creditworthiness of the organisation before offering a loan.

Demand for alternative financial services is rising rapidly and earlier this year Moula pledged to lend $250 million over the next 12 months.

The company has experienced strong growth over the past five years, processing over 20,000 business loan applications and growing its loan book by 124 per cent in 2019.

Green says the actual process of applying for a Moula business loan takes 7-10 minutes. It involves completing a fairly light touch application form, and granting approval for Moula to access the applicant’s business data. That’s generally achieved through an API into a data source, such as Xero, as well as additional insights gleaned from third party data sources such as credit bureaux.

Once the data is accessed, it is analysed, business risks assessed and a decision taken about the requested loan. “It’s typically a 24 to 48-hour underwriting approach, once we have all of the business data provided.

“A 10-minute application process and a 48-hour turnaround, which is well inside what would’ve typically felt like a 10 to 14-day process for a bank,” said Green.

According to BizData Director Nadav Rayman small business is now able to use its own data as evidence of the quality of the business. This accelerates the pace at which a business loan can be made available, and is a precursor to small business securing access to other innovative services such as Moula Pay which allows approved business to be paid up front for invoices while customers benefit from three months, no interest, no repayments opportunity.

Rachel Bondi, Chief Partner Officer, Microsoft Australia, stressed the transformative impact of Moula, noting; “This really democratises lending for small business – regardless of your size you can get what you need, when you need, as long as you have the data to prove your creditworthiness.

“As Australia heads into the Open Banking era we expect that the combination of data and AI will lead to genuine transformation across the financial services sector and improved outcomes and experiences for customers. Moula is a great pioneer in that regard.”

Once an SME has completed the online application and provided Moula access to its data, a machine learning decision model developed using Scikit-Learn machine learning algorithms that are accessed through Azure Machine Learning Service, provide a prediction of the probability (or risk) of the SME being able to pay back its loan.

Moula has been able to refine that prediction by using historical data from past applicants to train the model.

Moula is also leveraging App Services for .net, app service containers for Linux, Functions, Azure SQL database, ServiceBus, CosmosDB, Azure Documents, Data Factory, Key Vault, and Azure VMs.

To ensure the security of SME data Moula deploys multiple layers of defence and takes advantage of Azure security features such as Advanced Data Security and Dynamic Data Masking.

It also has documents and files stored in private storage accounts accessible via its own application and uses KeyVault to ensure there are no naked connection strings or access keys in any deployed applications. It also employs IP whitelisting and subnets to restrict access to some applications and VMs.