Augmented, not Artificial Intelligence: How lending tech makes underwriters’ jobs easier, not obsolete

Should risk professionals in the lending industry fear AI?

We look at whether the threat is real, the potential for better decisions around risk, how this fits with the culture in our sector, and the potential benefits for lenders and their underwriters.

Almost 100 years ago, British economist John Maynard Keynes predicted that, by 2030, we’d all be working a 15-hour week, with the bulk of our financial problems solved through exponential increases in tech-fuelled productivity.

Drawing on the realities of the Depression era, he also wrote about the fear of 'technological unemployment' – where technology destroys jobs faster than new uses can be found for labour – an issue as relevant today as it was in 1930.

While Keynes considered this to be a transition to the next phase in societal development, it’s very painful for the individuals affected by it, even temporarily.

Since technology started transforming our society, people have been caught between these two opposing forces: the fascination, optimism and promise of tech-driven growth, and the fear that it might make us obsolete.

The need for purpose

Even when freed from drudgery and financial necessity, people still want to work, as it satisfies their need for purpose, variety, significance, growth, contribution, connection, and a position in society.

While we’re nowhere near working the 3-hour day that Keynes predicted, tech that supports automation is a reality in most business sectors, including lending. The main issue is that its capabilities are misunderstood, generating resistance and delaying change that can improve profitability for everyone.

The media sometimes obsesses over cases of job loss following digitisation, and we want to offer a more realistic perspective.

Technology doesn’t eliminate jobs. It augments them.

Better decision making

In the business lending industry, ‘technology’ is all about information and using it for better decision making, such as approving loans faster and reducing the likelihood of bad debt.

It serves the industry to see the practical reality behind the hype, and tap into this potential for better decision making.

Our 2021 Corporate Banking Predictions Report included examples from leaders around the globe, and there’s a lot to draw on. Joana Negrao of HSBC UK’s Trade and Receivables Finance said:

With the advances seen in technology…the industry will respond with propositions that are better suited to customer needs and that reduce the complexity in accessing this type of financing.”

Danette Copestake, IT Director of Wyelands Bank, said:

Banks will be able to enhance performance and use real-time data to inform trading decisions or check for fraud.”

Allica Bank’s chief product and strategy officer, Conrad Ford, said:

We’ll see a further acceleration of the trend towards live insight such as Open Banking and ERP data to drive better loan decisioning and monitoring.”

Let’s take a look at the digitisation hype cycle of AI and the genuine culture shift it’s prompting. We’ll explore how augmented intelligence inspires more effective, more accurate decision-making that reduces credit risk and how it works for lenders at different stages of digital maturity.

We are in a (temporary) AI hype cycle

It’s essential to see past the hype. If we don’t correctly understand what AI can do today, we might easily make wrong assumptions of what it will do tomorrow.

Currently, most companies use so-called narrow AI, which focuses on very specific, narrow tasks with precise definitions. It’s used, for example, for internet searches and recognising images and speech. Sometimes it can outperform people in terms of efficiency, but in very specialised use cases and with clearly observable limitations.

Where any nuance is required, such as the complex, layered decision-making needed for business lending, AI won’t replace people anytime soon, and that understanding is beginning to be recognised.

But AI can definitely help.

AI is really good at processing and making sense of data. Through government mandates such as open banking, and the power of internet connectivity in general, banking and accounting data is opening up.

Open banking means that any person or company with a bank account should be able to provide any lender access to it. AI and non-AI analysis yields valuable information about how that account is performing. They can also bring in data from other sources, such as digital accounting systems – and because of their speed, they can keep the analysis up to date, even real time.

AI is better thought of as augmented intelligence that assembles information and enables people and teams to make better decisions, much faster.

Underwriters can apply greater sophistication to their lending decisions because the AI within the lending platform supports risk models updated for the digital age, based on more data and a richer context around the borrower’s situation.

What lending technology or lendtech can reliably do is enable banks, alternative lenders, and their underwriters to make better decisions across a broader range of businesses and assets, boosting their revenue, the clients’ business, and the overall economy.

The risk management culture needs an update

A lender treads a fine line between being the borrower’s friend (“Have some money!”) and being their policeman (“…but you have to stick to my rules”).

Lenders draw this line in different places – they have different risk appetites – and so can serve the market in different ways.

Bank managers of old knew their customers profoundly, from their financial position to their family situation and social standing, even if that wasn’t called “data” at the time. They had deep knowledge about potential borrowers, and they took the risk decisions.

Introducing computers to risk evaluations changed that, and underwriters lost agency. When lending moved from personal relationships to automated decisions, the risk models shifted from individual knowledge and gut instinct to a calculation based on the data accessible to the computer, which could only stick to the rules. If the computer system said “no”, there was nothing the lender’s staff could do.

To this day, most risk assessments follow the same path as they did in the 1990s, when technology first touched financial services companies:

• the risk manager receives a credit application
• they collect auxiliary information
• a risk model spreadsheet scores the application
• the underwriters make a decision based on this score and their experience

This process – as well as many others – is deeply embedded into lenders’ systems. It may not have changed – but the data available to it certainly has. Far more data is available, and much of it is real time. This means risk assessments can be vastly more sophisticated, enabling lending in a much wider set of circumstances, as well as much faster.

Data gives banks and alternative credit providers the means to make lending much more personalised, and do it at scale. Augmented intelligence tells credit managers more about who they’re lending to and, more importantly, what they're lending against (tangible or intangible assets alike) throughout the customer lifecycle – they can react to events, or triggers, as they’re revealed by the data.

Tangible assets are anything with a residual value.

For risk managers, this allows a shift from “Who am I lending to?” to “What am I lending against?”.

Redesigning internal processes around the capabilities of lending technology is the cornerstone of a powerful change that brings risk models up to speed with what both lenders and borrowers need.

Our work at Trade Ledger is focused on enabling this by harnessing technology to provide the data, context, forecasts, automation, and workflows that credit managers need to satisfy their goals and serve their customers. Our platform provides augmented decision-making for underwriters and risk analysts, which continuously improves through – and alongside – their work.

How an updated risk policy pans out in practice

When they get a loan application, some credit providers today still assess the risk of the client, even if an asset is available as security. This is arguably the wrong risk focus because, if the borrower can’t pay, the lender can rely on the residual value of the asset(s) to repay the debt.

What if no tangible assets are available, which is increasingly common as the service and digital sectors grow? Invoice financing offers an alternative. Something has been created, and a customer found for it, and an invoice raised. The lender just needs to know that the thing that’s been created can be sold for the value of the loan. And even if the borrower is wound up, the lender ranks ahead of most other creditors. Depending on their risk appetite or risk policy and regulatory requirements, an invoice finance lender may be comfortable lending say 80% or 40% of the face value of the invoices. Invoice financing is fundamental for building the capacity to lend funds to businesses with increasingly intangible assets.

It’s also possible for lenders to build up a picture of the risk associated with a particular borrower over time, or of a particular sector, and use this information to refine their risk policies.

Or by looking at the sales ledger as an asset, the lender gets a more accurate, up to date view of the borrower’s situation. They can use predictive analytics to assess how the business and its needs might evolve, based on historical data on similar companies. At some point the borrower might introduce new activity such as new products, or import and export, opening new opportunities for the lender.

Every lender should be able to lend against a sales ledger.

Invoice finance enables forecasts to be based on payment history, type of customers the borrower has, etc. Augmented intelligence uses the new data flowing in to learn and update its analysis so that subsequent recommendations and decision support are up to date and increasingly accurate.

Augmented intelligence can also identify patterns and compare new loan applicants to historical data. AI-infused solutions can guide underwriters, enabling them to make decisions faster and based on better reasoning, by focusing on the realisable assets rather than the borrower.

How Augmented Intelligence learns

A bank or financial services company needs to be able to satisfy the relevant regulator, such as the Prudential Regulation Authority (PRA) in the UK, that it can recover funds if necessary. Life is easier for risk managers and underwriters if they have a system that collects together all the relevant data. For example, the Trade Ledger lending platform can assemble the borrower’s sales ledger including dilutions, open banking data, data from credit reference agencies and the Australian Securities and Investments Commission or Companies House in the UK, and even utility data that shows the borrower is using the expected amount of water and electricity for a business of their type.

The Trade Ledger platform might suggest a lending rate of 80% for invoice finance based on this evidence; all the underwriters need to do is use this consolidated view of information, and then accept or reject the application. The augmented intelligence learns from them, this decision, and how the decision played out over time.

If the debtor (the borrower’s customer) fails to repay an amount within a certain timescale, the platform can recommend increases or decreases to the facility, based on the strength of the collateral, which is the invoices.

AI and digitisation work for every level of technology maturity

Our experience at Trade Ledger is that augmented intelligence and digitisation bring benefits for lenders and their underwriters at every stage of technological maturity, whether they’re working from paper and spreadsheets, or have already put in place some digital transformation.

There’s a lot we do with data. We help credit providers get information through invoice finance even when the debtor’s system isn’t digital or cloud-based. The platform pulls data from an applicant’s accounting package or even a spreadsheet, and combines it with historical information specific to their sector and rate the quality of the applicant’s customers.

These are just some of the ways we help companies make prompt, informed, and confident lending decisions and manage their credit risk.

How do we do it? Our lending platform enables data-driven lending. It gives lenders a simpler way of working, across origination, decisioning, fulfilment and management. It brings together all the data they need throughout the customer lifecycle, provides analysis, and shows at a glance the information they need to take good decisions.

The platform can present real-time information (from both public and private data streams) to underwriters so they can make decisions based on the key attributes of the borrower.

Massive payout

There is a massive potential payout for financial services companies willing to go beyond the AI hype cycle and embrace data-driven lending.

Widespread access to data can equalise the lending landscape between owners of tangible and intangible assets because it supports a broader and thus more accurate understanding of their specific business contexts.

Financial services companies can improve their risk models, processes, and even their business models, and improve their competitive position through excellent customer service.

It all starts with choosing to gradually adopt lendtech infrastructure and upgrading internal processes. Those willing to try it stand to reap significant rewards for their efforts.

When taking a look at both your business and your clients’ context, you should ask:

How much data and accurate risk management can I get out of something that is not digital?

If you’re unsure where and how you can start with data-driven lending for robust growth, get in touch with the Trade Ledger team.

We’d be happy to help you capitalise on the power of augmented intelligence.

‍