Every lender has AI now, so where do you win?

You’ve been eyeing your neighbor’s back deck for months, the one with the string lights and the built-in grill. You priced one out for your own house, nearly choked on the quote and thought, how hard could it be? So you watched a few dozen YouTube videos, bought a new drill, and spent half the summer driving to the hardware store on what you swore would be your last run. Your kids are annoyed you keep skipping the pool, and your partner is mad about the lumber pile blocking the garage. You’ve also made one memorable trip to urgent care after your thumb lost an argument with a hammer. Every relaxing weekend you had planned goes into becoming an expert on… joists. Congratulations, the deck looks… good enough! Could it have been better? Yes. Could a crew who builds decks for a living have knocked it out in a weekend? Also, yes. And you’d still have your summer, a happy partner, and an unsplinted thumb.

“Just because you can doesn’t mean you should” turns out to be a very expensive lesson, and lenders are learning it right now with AI. The models are cheap and the how-to videos are endless, so almost any team can build something that looks impressive at first glance. The harder question is what’s worth building.

AI features are table stakes now

Taking in an application used to mean 15 to 30 minutes of typing and submitting. Now, thanks to AI, it happens instantly with few mistakes. It’s a real win for your team, but unfortunately your competitors have the same AI advantage now. In McKinsey’s 2025 State of AI survey, 88% of respondents said their organizations regularly use AI in at least one business function. The tools keep getting cheaper, too. Stanford’s 2025 AI Index found the cost of using a GPT-3.5-level model dropped from $20 to about 7 cents for the same workload. When everyone can provide the same speed, why does a vendor send you the deal? When every lender has the same AI, where do you win?

Your data is your advantage

Even if your competitor licenses the same model or software you’re using, you have something they don’t. Your data! You have your deal history, losses and notes, so you know which vendors will send you a clean file and which ones will send you on a wild goose chase. The same goes for the credit box you’ve spent years fine-tuning and the verticals you know like the back of your hand. Let’s not forget the holy grail, your vendor relationships. Once everyone has the same AI, the lender whose AI can see and best use that data gets the edge.

So where do you start? First, make your data reachable so AI can read it. From there, every AI idea lands in one of three buckets.

1. Build it yourself when it’s part of how you win deals.
2. Buy it from a specialist when every lender needs it and nobody picks you for it.
3. Switch on the AI already included in your platform for everyday work.

Start by making your data reachable

If AI can’t read your data, nothing else in this article is going to work for you. An API works like a waiter at a restaurant. It carries a request from one system to another, like “send me this customer’s payment history,” and brings back exactly what was asked for. MCP, short for Model Context Protocol, is a newer standard built for AI. It gives AI agents and assistants one common set of rules for working with your system of record. An MCP server is what puts those rules in place for your system, so AI can pull a deal file or update a status without anyone exporting a thing.

Not sure whether your data is reachable? Try picking an easy portfolio question, like “Which equipment types had the most late payments last year?” Ask your ops or IT lead how they’d get you the answer. If an AI assistant connected to your system can answer it in seconds, you’re in good shape. If the answer involves exporting a spreadsheet and losing an hour to pivot tables, start there. Reachable data also lets you switch tools later without starting over.

Decide what to switch on, buy or build

Before your team builds anything, ask yourself one question. Would a customer or vendor notice if you did this the same way as everyone else? If the answer is no, remember the deck. Just because you can build it doesn’t mean you should, especially when someone has already built it for you.

Switch on the AI you already have. Intake, document sorting, and data entry come with plenty of LOS and lease management systems today. If you have those features available, turn them on before you build anything new.

Buy the table stakes. KYB, fraud checks, OFAC screening, and bank statement analysis matter to every lender, but vendors aren’t sending you deals because you built your own OFAC check. Thankfully, specialist integrations run these checks all day at scale.

Build what makes you different. McKinsey found 80% of organizations use AI to save time and money, but the companies making the most money from it also use AI to grow. Growth comes from the places where you underwrite, price, or go to market differently, so save your time, money, and patience for those. For example, a broker might use AI to match each deal with the right lender. Matching deals to lenders is that company’s whole business, so using AI for that function ties straight to their strategy. For a lessor, the same thinking might point to AI trained on your own credit box or a broker scoring model built from years of your own funding and loss data.

The cost of building shows up later

AI can get you 80% of a feature fast when you need it. The last 20%, meaning all the edge cases and exceptions, takes far longer. After launch, the feature needs you to take care of it forever, so when models update, rules change, or something breaks, guess who gets the Friday afternoon email about it? S&P Global Market Intelligence found the average organization scrapped 46% of its AI proofs of concept before they reached production. Yes, plenty of those teams could build the demo, but the question was answered about whether they should.

For most smaller independents, building means setting up general AI tools against your own data, and one or two custom tools is plenty.

Where to start Monday

Start small. The goal is gaining a clear idea of what makes vendors call you first, and you can get there one decision at a time. Pick the one thing vendors would miss most if you did it like everyone else and make sure AI can read the data behind it. For everything else, switch it on or buy it, and bring the checklist below to your next LOS or AI renewal. Just because you can build it all doesn’t mean you should, so leave the decks to the crews who build them every day.