Commercial Lending in an Hour

Note: This is a transcript of my YouTube walk on Commercial Lending.

—

Good morning.

Today’s walk is going to be on commercial banking. Not the kind of business that’s hard to explain. You’ll get it pretty quickly. But it is the kind of business that’s surprisingly hard to transform, especially through technology…. which is my intent on this walk: to talk about commercial banking through the lens of generative AI… within the context of the current lifecycle of a commercial loan.

I’ll say from the get-go that commercial banking is a relationship business and tech is usually an efficiency multiplier, not a relationship multiplier. If anything, digital transformation tends to broaden, not deepen relationships… which is my way of saying that there’s a real risk of tech negatively impacting commercial relationships.

An everyday banking example of that would be: every old-school bank is trying desperately to redirect you from their banking center– those buildings with the tellers– to their mobile apps. They’re doing it to become more efficient. Tech as efficiency multiplier. To save on rent and paying tellers. And every step the larger banks take toward eliminating their physical footprint– well, they lose the human connection in banking and consequently, their ability to differentiate from digital-first, digital-only banks. So– as I said– tech (without being more mindful, and human-connection-centric) is usually an efficiency multiplier, not a relationship multiplier.

Ok… so if you’re not familiar with commercial lending, let’s do a quick overview… even if it’s painfully obvious.

Commercial lending is the process of making loans to businesses (commercial banking) rather than individuals (consumer banking). This type of lending allows businesses to borrow money to finance anything from purchasing inventory to expanding operations to investing in new equipment.

Commercial loans typically have higher interest rates compared to consumer loans due to the higher risk associated with lending to businesses. That doesn’t sound right intuitively because an individual has less money than a business. But remember that businesses borrow exponentially more money and more regularly than consumers so they have a higher likelihood of defaulting on their loans… because they’re more sensitive to economic downturns, market fluctuations, changes in industry trends, poor management, competition from digital-first startups.

Also… and again, this might sound counterintuitive but… businesses may have less stable income compared to individuals— especially if you look at the relative size of loans (you and me— think mortgage) (a business… think factory, or robotics or payroll… “capital intensive”… which is another way of saying “way more expensive than our mortgage”).

And compare the regularity of income (you and me— we get a regular paycheck) (a business — their levels of cash-on-hand can vary greatly from day to day, month to month, quarter to quarter).

So a bank correctly views business loans as higher risk— they understand that businesses (compared to Mom and Pop taking out a loan for a place to live).. businesses are less likely to meet their loan obligations. And that translates to higher interest rates on commercial loans to compensate for this increased risk.

Look. That’s part of it. The other part is banks aren’t stupid — they understand the competitive landscape for money– for loans– and they realize that businesses need the loans (capital N need). And when you have that kind of demand, you can charge a premium for it.

Yes, a business's creditworthiness is a factor and we’ll get into that when we talk lifecycle. The bank’s going to want to do a close review of the company’s financial statements— their future projections. And, depending on the size of the loan, the bank might even want to review the company’s business plan for that money. They might want to insist on a certain level of maturity in the company’s financial management. All of that helps reduce the risk to the bank— and that process increases the company’s chances of being approved for the loan.

But there will still be a premium on that loan. So commercial banking is lucrative.

How lucrative? This year, the Global Commercial Banking Market is estimated to be $2.24 trillion (with a T) and is expected to reach $3.81 trillion by 2028. The business world has an unquenchable thirst for loans for everything from financing inventory to expanding operations. So… high demand… which– as I mentioned 20 feet back– allows banks to charge a premium for providing these loans.

One quick definition before we jump. What’s a credit facility and is it different than a commercial loan? A credit facility is a type of commercial loan that provides the borrower with a revolving line of credit. This means that the business can borrow funds, repay them, and borrow again up to a predetermined credit limit… without the hassle of all the bureaucratic steps of the single loan process happening every time you need another loan. A credit facility is more flexible than a traditional commercial loan because it allows the borrower to spend the funds as needed and only pay interest on the amount borrowed. In contrast, a traditional commercial loan provides the borrower with a lump sum of money that must be repaid in regular installments over a fixed period of time. Once repaid, that deal’s done. You want another deal. Fill out the form again.

A credit facility on the other hand is like your parent’s credit card. You can use it. You have to pay the bill. But if you do– pay the bill– you can use it again. Credit facility.

Alright. Let’s talk operational lifecycle.

At a bank, the lifecycle of a commercial loan is typically divided into several stages. The first stage is the application process, where the business submits a loan application with relevant financial information and documents. Once the application is received, the lender reviews the business's creditworthiness and evaluates the potential risk involved in lending to the business.

If the application is approved, the loan moves on to the next stage, which is the underwriting process. During underwriting, the lender performs a more detailed analysis of the business's financial statements, cash flow, collateral– anything really, that can shed light on the risk to the money being loaned. And all that’s done– not just to say yes or no to the loan (the next step)– but to factor in thoughtful terms and conditions for that loan. Thoughtful to the bank and its legion of lawyers. This stage also involves requesting additional documentation and conducting site visits or property appraisals… anything that will give the bank peace of mind about the risk they have to take.

The next stage in the commercial lending process is what I hinted at: the loan approval stage. Here, the lender determines whether or not to approve the loan based on the information gathered during underwriting. If approved, the lender will provide the borrower with a loan agreement– a contract– that outlines the terms and conditions of the loan.

Once the loan agreement is signed, the funds are disbursed to the borrower, marking the funding stage. The borrower can then use the funds for the intended purpose, whether it's purchasing inventory, expanding operations, or investing in new equipment… whatever.

And… like with your mortgage, throughout the life of the loan, the borrower is responsible for making regular payments according to the agreed-upon schedule.

There is a lot of complexity in one step of that process– the underwriting step– and that’s mostly because underwriting is human-heavy… operationally. The process for gathering data about the business and the loan is standardized and mostly automated… but depending on which human underwriter (or underwriting team) gets that workflow step… the ultimate thumbs up or thumbs down on the loan is dependent on human judgment.

That human (or team) thoroughly evaluates the borrower's financial information to assess creditworthiness and determine suitable terms and conditions for the loan. That human (or team) analyzes financial statements, cash flow, collateral, and other factors to make an informed decision. It’s even up to that human’s discretion to request additional documentation… which again, they would review. Human underwriters– guided by thoughtful policies and controls and a ton of training– ensure that the bank mitigates risk and provides loans to businesses with a high likelihood of repayment.

Let’s call that whole step– underwriting– more art than science. You definitely risk the hungry judge effect… the idea that judges– in the courts at least– are more inclined to be lenient after a meal and more severe before the meal break. And hopefully, we’ll come back to this.

Suffice it to say: to run an effective shop, a commercial bank needs to establish clear processes and procedures. This includes having a dedicated team that handles commercial loan applications– a separate team for underwriting, approval, and disbursement. The teams should have a solid understanding of the bank's lending policies and guidelines… no duh. And it's important for the bank to stay updated on industry trends and regulations that may impact commercial lending. No double-duh.

Ok. Ops aside. What do businesses look for in a banking partner for commercial loans?

Let’s explore the obvious. Businesses look for:

* Experience and Expertise: They prefer banks that have a demonstrated track record and expertise in commercial lending. But they really want a banking partner that understands their industry, their history as a company, their specific needs… so the bank can use that information to tailor a custom solution for the business. That’s why relationships are so important– the deeper the better. The CFO of a company might say “I’ve known Banker Bob for years and trust him”... but what underlies that statement… is “Banker Bob knows my business. He knows my constraints. And he trusts me.” Can a robo-lender earn that kind of trust. Maybe… if it’s designed up. What else?

* Competitive Interest Rates and Terms: Businesses– like all of us– are looking to save a buck. They’re looking for flexibility on repayment terms and because everyone likes to feel special, favorable loan conditions. They want to ensure that they can afford the loan and that it aligns with their company’s financial goals. Banker Bob can do that because there’s history there (between him and the CFO). A deep business relationship. Can a robo-lender do that? Maybe… if it’s designed up. What else?

* Quick and Efficient Loan Processing: No one wants to wait forever. So businesses value banks that have streamlined and efficient loan processing systems. If you’re going to wait for the Star Wars ride at Disney, you don’t want to wait for 5 hours. The typical business loan from a bank can take weeks to process or months. Some banks differentiate by demonstrating that they can fund within a week of a business submitting an application. This is where a robo-lender already shines. Products like Rocket Mortgage… which is a little bit of marketing misdirection because it’s their application process that’s fast. While their underwriting still takes up to a week. The tech challenge is to automate underwriting– to evolve from a human-centric, SME-centric (SME is subject-matter-expert) a human-centric, SME-centric process that takes hours and days– to an AI or algo evaluates risk within seconds, that forms terms and conditions to address those risks within seconds. But I’m getting ahead of myself. What else do businesses look for in a banking partner for commercial loans?

* Responsive Customer Service: Does a human pick up a phone? Is that human useful once they’ve picked up the phone? Everyone appreciates responsiveness and personalized customer service. And the more complex the product, the greater the need that the person picking up the call is a SME. In consumer banking, the call center gets asked questions like “What’s my balance?” Or “What’s this charge?” In commercial banking, the questions are harder to answer. And in today’s contract-centric world– where each loan is a contract and each contract is a snowflake that’s been customized for a specific client (or more accurately, customized to address specific risks with that deal), the SME– that poor person picking up the call– would need to read that contract in real-time– or to quickly isolate the parts of the contract that they need to read in real-time… making it hard for them to answer what are essentially legal questions in real-time. Lawyers can’t even do that. Or have been trained not to do that. Again, getting ahead of myself but the industry needs to move from contract-centric to data-centric… from giant blobs of legal text to something the average call center employee can use in real-time… because without that you have to direct all calls to the deal team and they’re hopefully already busy working on the next 10 deals. So rerouting questions to the deal team or to lawyers doesn’t scale. Call centers scale. But the products they support need to be simplified with support in mind. Because just picking up a call quickly doesn’t mean great customer service. What else?

Note that I haven’t explicitly called it out with everything I’ve touched on in the last 5 minutes but we’ve covered relationship building… or why relationship building is important. Who doesn’t value their lending partner taking the time to understand their business? We’ve covered why industry knowledge is important. Who doesn’t value a partner who has a deep understanding of their industry and can provide valuable insights and connections? We can infer from those two things, that a deep relationship with a lending partner who really understands your space can lead to them offering strategic advice and connecting you with other professionals or resources that can enhance their business growth. Think private equity. Think relationship centricity.

But let’s continue. What else do businesses look for in a banking partner for commercial loans? I think I touched on flexibility already but…

* Flexible Financing Options: cookie-cutter options are great if you’re a cookie. Humans prefer banking partners that offer a variety of financing options to meet their specific needs. Whether it's a term loan, line of credit, or equipment financing, businesses need the flexibility to choose the most suitable option. What else?

* Digital Banking Capabilities: I know that sounds obvious but businesses aren’t consumers. There isn’t a CFO out there thinking “I should be able to use my personal phone to pay my company’s bills.” But every CFO is thinking “Why are my systems not fully integrated with my banking partners' systems?” Yes, they offer APIs aaaand… I don’t have a bunch of developers to actually use them. Don’t get me started– I’m talking to myself here– about the arrogance of offering APIs as a solution. The arrogance of banks offering “convenient online and mobile banking platforms” when the business owner doesn’t live their digital life in the bank’s properties. Anywho… if I’m just listing expectations… digital is in there because, in theory, it allows businesses to easily manage their accounts, make payments, and access important financial information. Or so banks keep telling themselves. What else?

* Stability, Resiliency: A strong reputation both in the financial sense and in their tech. Think reliable, trustworthy, 24/7/365… throughout the duration of the process and the loan. And I could go on forever so I’ll do one last one:

* An Ability to Meet Funding Needs: If you’re Giantco, you’re probably not going to browse to fintech.com for your massive lending needs. You’ll flirt with them to get better rates from the big guys. But that’s just good negotiation. The point here is that different-sized businesses look for different-sized banking partners…. ones that have the capacity to meet their funding needs. Whether it's a small loan or a large financing requirement, businesses want to ensure that the bank has the resources, the scale, and capabilities to provide the juice. Before I jump off this point, it’s worth noting (and pretty well-known in banking circles) that small and medium-sized enterprises have a relatively disengaged relationship with their banks. Accenture published some research that said that 47% of small and medium-sized businesses believe that banks don’t try to understand their challenges. Only 9% are comfortable that their current bank meets all their needs. That’s a huge gap... a huge market opportunity for entrepreneurs or intrapreneurs.

Anyway… I ended on that because large lenders have an economic advantage over smaller or nimbler lenders… which is one explanation for why startups aren’t yet eating commercial lender’s lunch. Large lenders can and do offer lower rates because they can and do take advantage of economies of scale. They can and do get volume discounts on money. They can and do access cheaper or more abundant capital.

That’s important to remember because the value of a digital-first, fully automated, easy process isn’t yet enough to change business behavior. Another way to think about it: If you’re a big, slow-moving beast of a business, you probably don’t mind partnering with a big, slow-moving beast of a bank. And the final way to think about it: relationships are always more important than transactions. It’s not just that Bob the Banker has an advantage. Bob is the advantage.

Also worth noting– and probably a 30-second chapter on its own… two words: trade finance. If you don’t know what that means, trade finance is banking jargon that describes different strategies employed to make international trade easier. Digital payments for instance. It’s financing… for trade– Trade Finance. Think domestic and international trade transactions. Think: services for the sellers and buyers of goods.

Anywho… Trade Finance isn’t exactly a hotbed of innovation. But customers have come to expect more and better… startups (or what bankers call nontraditional competitors) aren’t yet putting the traditional players out of business… but… Accenture put out this report last year called “Finding Your Competitive Advantage in Trade Finance” or something like that and in it, they talked about how 40+% of North American banking customers said fintechs were a workable alternative to bank-provided payment services. If you're in the Treasury management business, that should scare you. That same report listed all the reasons that survey respondents cited about why they might choose a fintech over a bank: cost… was number one... high 50s in terms of the percentage of respondents who mentioned cost as the reason they're looking past banks), digital capability (30-odd%), ease of use (low 50ish%). ease of integration (40%ish).

Those are big numbers… that have lit the fire under bankers’ asses.

Every commercial bank is under pressure to address the gaping digital hole in trade finance. And the more entrants and the slicker the tech of each new competitor, the more trade finance (as a function) is stepping up their investments in technology (hopefully), and stepping up their product offerings (hopefully).

That’s my way of saying it’s the right time to have this next part of the walk.

Ok… let’s get to Tech and the art of the possible or

How can AI be leveraged to radically improve the commercial loan experience?

Think of generative AI as the mother of all Google searches across all the documents you already store. But it's also a time machine. It can gather relevant information from applicants even before they apply for future loans.

AI can reduce the need for manual data entry (once they do apply) streamlining the application process… again… even before the client applies for the loan.

Stop. Let me rewind. It might have made sense (for a walk like this) to start the application of AI when the client expresses an interest in a commercial loan. But that’s yesterday’s thinking… It’s too late.

Step 1 should be to leverage AI for prospecting and sales… 1) for organic growth with the clients you already have and 2) for finding new prospects.

So organic growth first… and if you don’t know what organic growth means… it’s a fancy way of saying “Let’s try to get more money from the clients we already have.” It’s sometimes called “deepening your share of wallet”… the banking equivalent of “Do you want fries with that burger.” Because when your client’s wallet is already open, that’s a good time to sell them more… or to sell them adjacent products…. fries for every burger. That’s organic growth.

So… if your bank already has 10K clients— whether they’ve taken commercial loans out from you in the past or not— doesn’t matter… you can feed an LLM— a large language model (LLM)— an AI platform… you can feed their profiles… your client’s profiles. It might even be as easy as feeding the LLM their previous contracts– those massive, complex Word documents filled with legal jargon and fee schedules.

You’ll also want to feed the LLM your risk boundaries (also in the form of a Word document)… and then craft a prompt that synthesizes the data in those documents into a prospect list…. A list of who to call for sales… There’s little risk of the AI hallucinating because you’re feeding it the unstructured text that you want it to use (i.e., the client’s previous contracts, their previous loan applications, their past and current financial information, any supporting documents they provided earlier, even notes from your own underwriters when they were evaluating previous deals). And you’re allowing the AI to infer what’s risky about the client from that unstructured data.

I laugh because the ease with which you can do that is going to create a nightmare for making tech investments in your legacy applications. You shouldn’t have a million contracts in your contracts database. You should move that whole contract writing process from its analog “let’s just write a doc” process to a “let’s capture all the data and rules for creating a doc.” But once your business sees an LLM’s ability to infer from that unstructured contract… they’ll be hesitant to change their analog habits. They’ll blame it on how expensive and disruptive the change will be… but it’s just an excuse to keep the processes that they grew up with.

So… LLMs unintentionally will increase technical debt in every organization that employs them.

But let’s keep going. Where’d we stop?

Generative AI has now produced a list of clients to call to offer new loans proactively… before they reach out to say they need one. It almost doesn’t matter if the AI’s suggestions are correct because it lights a fire under the banker’s ass… the salesperson is forced to reach out and… if nothing else, talk about how cool it is that generative AI has asked the banker to call that client. Relationship deepening…. and the best kind because the client will be both spooked and impressed.

This is a rare example of how tech can be designed to be both an efficiency multiplier (sales enablement improves) AND a relationship multiplier (tech facilitates positive human connection, deepening relationships). It gives two people a reason to talk to each other, to engage one another– a reason that’s important to both.

That’s step .5– we’re not even at step 1 yet– and it’s a great example of the art of the possible. Let’s go to step 1.

In the old world, if the client was ready to start the loan process, they’d have to fill out all kinds of forms. No need in the age of AI. The same kind of prompt engineering that got you a list of prospects can be leveraged to produce the actual applications that need to be filled out. How? By feeding the LLM the unstructured garbage in your contracts repository and your CRM systems. Engineer a prompt that asks for the data in each contract. In other words, reverse engineer the thoughtfulness that should have (but didn’t) transform your rickety contracts management platform into a data management platform. For the data nerds watching, that’s also how you populate and/or validate your client master (your programmatically-available golden source for client data). To your relationship managers’ delight, the client’s role goes from a tedious data entry specialist to a more sophisticated data reviewer.

Next step: AI can re-analyze the client’s data to assess their creditworthiness more accurately and efficiently. Not through generative AI but through custom algos and predictive modeling, banks can make faster and more informed decisions about loan approvals, minimizing the time it takes for applicants to receive a response. Down from weeks or months to seconds.

So in that last step where the client is just making sure that all the data in the form is right… once they’re sure it is and they hit the proverbial submit button, the underwriting process feels done. Why? Because it was done before the deal was put in front of the client.

By leveraging this proactive model, banks are essentially pulling the underwriting process of commercial loans to happen before anything is signed. The banker doesn’t need to engage underwriting because the AI algos have already analyzed vast amounts of data (those financial statements I mentioned, current market trends, current industry benchmarks) the model has already assessed the risk associated with a loan. The human underwriter’s job transforms from “keeper of tribal knowledge” into the memorializer of that same knowledge; from judge to law-maker. The function upgrades from risk management to risk model management.

What’s cool is that underwriting stops being the operational bottleneck because the bank has in essence already made a more informed decision about interest rates… and loan terms… and collateral requirements… even before their client has come to them hat in hand.

And what’s super-cool is that an LLM’s chatbot capabilities (think ChatGPT) (on all those documents we just fed it) just gave your customers access to better data and advisory insights. It can be an enabling tool that lets’ companies slice and dice their data, giving both lender and lendee a competitive edge. Heck, LLMs can even assist in identifying potential fraud or suspicious activities, further mitigating risks for the bank. But security’s not my core competency so I’ll stop there.

Remember how at the beginning of the walk I talked about the difference between a commercial loan and a credit facility? Remember how… when a business takes out a single loan, if they want another one, they need to apply again and go through the hassle again. Remember how I talked about a credit facility being like your parent’s credit card? How you can use it, pay the bill, and use it again? Well, what I just outlined with the potential uses of AI… can turn every one-off commercial loan into a properly risk-managed credit facility.

And that deserves a hallelujah from every business person on the planet who has ever filled out a loan application… or worse, filled it out again.

Look. This walk isn’t about "improving the customer experience." This isn’t about increasing the efficiency of the bank's lending operations.

This is a complete reimagining of the end-to-end… without the constraints of having to first modernize all your legacy platforms.

For example, when it comes to disbursing the loan, the AI model has already automated the verification and documentation process. If you’re an engineer, the whole notion of real-time calculation or real-time processing is yesterday’s ambition.

Real-time with AI is too late.

The paperwork is already “complete and accurate” two steps earlier in the business workflow.

The chances of errors or missing information are resolved at step one of the data supply chain.

Will that lead to faster loan disbursement?

If anything, we need to put in controls so it doesn’t happen too fast.

And the best part… Just like AI would change the client’s role from a tedious data entry specialist to a more sophisticated data reviewer… AI can and will change every operations role on the bank’s side from a tedious manual reconciler to a more sophisticated decision scientist.

That’s a change that’s long overdue. Back-office operations need to enhance their employee experience and what I’ve talked about will necessarily revitalize the poor folks doing all the hard work. And it’ll attract fresh faces to the industry and to the function… all of whom will come with strong digital competencies.

That’s a pretty hopeful place to end the walk.

I hope you learned something.