Artificial Intelligence in Lending: Where Friction Disappears and Speed Begins

Ask any SME owner what they remember about applying for loans, and chances are it isn’t the outcome. It’s the wait. The back-and-forth over documents or the hassle of submitting an application and hoping for the best. 

For decades, this has simply been how commercial lending works. A business applies, an underwriter reviews, questions get raised, more documents get requested, and eventually a decision arrives. Each of these steps made sense when they were introduced. Together, they have created a process that often takes longer than the business problem it’s meant to solve. Artificial intelligence is starting to change that, not by replacing the fundamentals of good lending, but by removing the friction that the process involves. 

Lending Without the Friction

It’s worth pausing on why lending became so document-heavy and manual in the first place. Underwriters need information to assess risk, and lenders need to verify that information is accurate. Compliance requirements add further checks. None of this is unreasonable on its own. 

The problem is how these steps have typically been stitched together. Application intake, underwriting, credit decisioning, and servicing have often operated as separate stages, sometimes handled by different teams, on different systems, with information re-entered or re-verified at every stage. Friction wasn’t a deliberate design choice. It was simply what happened when each part of the process was built and managed independently. For a business waiting on a decision, none of this context matters. What matters is how long it takes to get the loan approval and how much effort it requires. 

What Changes When Data Moves in Real Time 

Traditional underwriting relies heavily on historical financial statements, often months old by the time they’re reviewed. This isn’t because lenders don’t value current information, but because gathering and processing it manually has always been slow. 

AI changes this equation by making it possible to work with real-time data at scale. Bank transactions, accounting records, and trading activity can be analysed as they happen, giving a far more current view of a business’s financial position than a set of year-old accounts ever could. This isn’t just a theory. Independent research estimates that better use of real-time data in lending decisions could be worth up to £570 million a year for UK SMEs. 

This matters because business circumstances change quickly. A company that looked financially stretched six months ago might be in a completely different position today. Real-time data allows lending decisions to reflect where a business actually stands, not where it stood when the last set of accounts was filed. 

Speed as a Byproduct, Not the Goal 

It’s tempting to talk about AI in lending purely in terms of speed, and speed is certainly part of the story. But faster decisions are really a byproduct of something more fundamental: a process with fewer unnecessary steps. 

When application data flows directly into underwriting, when underwriting connects directly to decisioning, and when decisioning connects directly to servicing, there’s simply less waiting built into the system. Decisions that previously required days due to the frequent transfer of a file between parties can now be made promptly, as the file no longer needs to change hands. 

This is the principle behind Pulse’s Unified Lending Interface (ULI), which brings application intake, underwriting, decisioning, and servicing together within a single ecosystem. Rather than treating these as separate stages connected by manual handoffs, ULI allows information to move through the lending lifecycle as one continuous process. 

Nucleus Commercial Finance has built its lending approach around this kind of connected infrastructure. As a lender, Nucleus has adopted embedded lending capabilities through its partnership with Pulse, with the aim of removing the operational drag that has traditionally separated a good application from a fast decision. 

Where Einstein aiDeal Fits In 

Connected infrastructure solves part of the problem. The other part is the assessment itself, which is where Pulse’s AI-driven underwriting engine, Einstein aiDeal, comes in. 

Einstein aiDeal is designed to assess applications using real-time data alongside credit criteria defined by the lender. The model doesn’t operate on a generic, one-size-fits-all view of risk. It applies the specific lending policies and risk appetite that Nucleus has set, just at a pace and scale that manual review can’t match. In practice, this means the majority of incoming applications, around 95%, can be assessed and decisioned in less than 45 seconds. 

What This Means for the Remaining Applications 

Automating the majority of decisions doesn’t mean removing human judgment from lending. It means freeing it up for where it’s needed most. Applications that fall outside standard criteria, that involve unusual circumstances, or that simply need a closer look still benefit from experienced underwriters. The difference is that those underwriters are no longer spending their time on the high volume of straightforward applications that didn’t need manual review in the first place. 

For Nucleus, this means underwriting capacity can be directed towards the deals where it adds the most value, while the bulk of applications move through the system at a speed that wouldn’t have been possible under a fully manual model. 

Why This Matters Beyond the Numbers 

It’s easy to talk about automation rates and decision times as efficiency metrics. For the businesses on the other end of these decisions, they translate into something more practical: being able to act on opportunities as they arise, rather than waiting for a process to catch up with them. 

A supplier offering a time-limited discount, an unexpected order that needs upfront funding, or a short-term cash flow gap before a large invoice is paid. These are the kinds of situations where a 45-second decision and a multi-day decision aren’t just different in degree. They can be the difference between an opportunity taken or missed. 

This is the shift that AI-driven underwriting represents for lending. Not a replacement of sound credit assessment, but a removal of the delays that sat around it. By bringing application intake, underwriting, decisioning, and servicing into a single connected process through ULI, and by using Einstein aiDeal to assess the majority of applications against real-time data and defined credit criteria, Nucleus is able to offer SMEs something that reactive, document-heavy lending processes have struggled to provide: a decision that keeps pace with the speed of business. 

Speak to us to learn more about our funding solutions.