Missed calls that cost you loans. Slow data lookups during live underwriting calls. Departments that don’t talk to each other without someone manually pushing things along.
These are the friction points that quietly bleed margin out of a lending business every single day.
After funding thousands of loans at Hard Money Bankers, I can tell you AI is now hitting all of those problems at once. I want to give you the real picture of what’s working right now, not what sounds good in theory. I co-host the Private Lenders’ Podcast and run the Hard Money Mastermind, and I hear from lenders every month about what’s actually moving the needle.
Key Takeaways
- AI tools now operate across every department in our lending business, not just marketing.
- Voice AI is converting about 80% of missed inbound calls into booked appointments.
- Conversation AI on our website is generating roughly 3 new leads per day in early testing.
- I built a custom loan origination platform using AI assistance in about a month.
- MCP server connections to The Mortgage Office and SFR Analytics are the newest and most powerful additions.
- Start small. Do one project. Then add to it. Don’t try to do five things at once.
In This Article
- AI for Marketing: Content, Ads, and Data
- AI for Sales and Lead Conversion
- AI for Loan Origination
- AI for Loan Servicing
- The Right Mindset for Implementing AI
- Frequently Asked Questions
AI for Marketing: Content, Ads, and Data
On the marketing side, AI is doing a lot of the heavy lifting. I use large language models to create content skeletons, and then my virtual assistants refine and build on them using knowledge bases I’ve put together. Those knowledge bases are built from FAQs, past podcast episodes, and website content.
The LLM isn’t writing the final product. It’s giving me a scaffold that a human then develops.
I want to be direct about this. Do NOT just take a ChatGPT or Claude output, copy-paste it onto your website, and call it done. I’ve heard from marketers who follow the search engine space closely that AI-generated content with no human refinement is being flagged.
Search engines are getting better at identifying low-quality, undifferentiated AI content and deprioritizing it in rankings.
Use AI as a resource for identifying the most relevant topics getting eyeballs right now, then expand on it with genuine human expertise. AI compresses lending timelines and marketing alike, but the human layer still matters.
Beyond content, I use AI-assisted tools for Google Ads split testing, Meta ad testing, and SEO keyword research. Shane, who handles a lot of my web marketing work, manages a lot of that. And one tool I almost left off this list is Opus Clips.
I use that to take longer videos and carve out short-form clips automatically. Saves a ton of time.
The SFR Analytics and Claude MCP Connection
This is the one I get most excited about right now. I connected SFR Analytics to Claude through what’s called an MCP server. If you’re not familiar with that term, just Google it.
I wasn’t either until a few months ago.
What it means practically is this. Instead of logging into SFR Analytics, running reports, exporting data, and then analyzing it somewhere else, I can just type into Claude: “What were all the investment transactions that closed in zip code 20743 last week? Who were the lenders?”
“Who were the borrowers? What was the average purchase price?” And it gives me the answer instantly.
I also use it for quick underwriting verification. If a borrower tells me on a call that they’re an experienced investor who’s done a lot of deals, I can check their entities against actual transaction records in seconds. Instead of bouncing between MLS, land records, and public records, I just type it into Claude through the SFR connector.
That’s a REAL time saver when you’re on a live call trying to make a fast decision.
For a bigger data pull, like everything that closed in a market over the past year, I still go directly to the SFR Analytics interface. But for quick, cross-referenced lookups, the MCP connection to Claude is genuinely useful. McKinsey AI banking value, and the MCP approach is one of the clearest examples of why.
AI for Sales and Lead Conversion
This is where things get particularly interesting. I started with voice AI, and that’s still running strong. When a call comes in and none of my loan officers pick up, it goes directly to a voice AI agent.
That agent’s only goal is to book an appointment. It doesn’t try to give rates. It doesn’t try to close the deal.
It just tries to get them on the calendar.
I’m converting about 80% of those missed calls into booked appointments. That’s the stat that surprised me most when I first saw it. The ones that don’t book still give me enough information, because I have the inbound phone number, and that gets routed to the originator whose number they called.
So even a failed booking still creates a contact record and a follow-up task.
If you want a deeper look at how I structure the front-end lead conversion process and follow-up sequencing, I’d point you to our breakdown of how to convert hard money leads from inquiry all the way to closed loan.
Conversation AI on the Website
I used to run a chatbot called Tidio on the Hard Money Bankers site. It worked on a branching logic model. I’d build out the branches manually based on what questions people might ask.
I collected about 6,500 contact records through it over the years, which was genuinely useful.
But I’ve now switched to Conversation AI inside GoHighLevel. The difference is significant. Instead of rigid branches, it’s a live AI agent powered by a knowledge base.
You feed it the same knowledge base you use for voice AI, give it the primary goal of booking an appointment, and it handles whatever comes up.
The interaction looks something like this. Someone lands on our site and the chatbot opens: “Hey, I’m Jason at Hard Money Bankers. Do you have a project you’re working on or any questions?” They ask if we lend in Maryland.
The AI confirms we do, asks for their name and number, and tries to get them on the calendar. If it gets a question it can’t answer, it’s transparent about that and redirects to booking a call. It doesn’t invent answers.
I’m about two weeks into this and already seeing roughly 3 new leads per day from the chatbot alone. I’m still connecting conversion pixels through Google Tag Manager, so I expect that number to get more accurate as I track it properly. But the early signal is good.
On the backend, when a perfect interaction happens, GoHighLevel creates a contact record, books a calendar appointment, captures name, phone number, and email, and routes it to the right originator.
When it’s a partial interaction, I at minimum have a phone number to work with. It’s not perfect yet, but it’s additive to what the human team was already doing.
AI for Loan Origination
This is the one that required the most custom work but has maybe the biggest day-to-day impact on my team. For years, I ran my origination entirely in Google Sheets. I liked it.
My team was trained on it. I’d tried other loan origination softwares but kept coming back to Google because it worked the way I worked.
The problem was that every new loan required creating a new Google Sheet. Every new month required a new pipeline tab. And the reporting was fragmented.
I could figure out where my loan volume was coming from, but I had to piece it together from multiple places.
Building an AI-Powered Hard Money Origination Platform
So I built my own. With AI assistance and my developer, Christian, I built a custom origination platform in about a month. I used my existing Google Sheet structure as the template and said: build something that works exactly like this, but as a proper pipeline tool with notifications and reporting in one dashboard.
Now when a loan moves from lead to term sheet to in-processing, my portfolio manager Tara gets automatically notified. No manual handoff. No one has to remember to send a message.
The pipeline moves, the notification fires, and she picks it up. That’s the kind of thing that seems basic but in practice eliminates a whole category of dropped balls.
The reporting is the other big win. I can now see in one place exactly what percentage of my loan volume came from brokers, repeat borrowers, Google Ads, and other channels.
Before, I had that data but it was spread across different tools. Now it’s in one dashboard. If I see that 15 to 20% of loan volume is coming from broker relationships, that tells me where to put more marketing dollars and attention.
This is the kind of data that drives real decisions about scaling loan volume.
I’ll be honest: if anyone looked at my software from the outside, they’d probably ask why it doesn’t do 10 other things. And the answer is because I don’t need it to do those things. It’s built for the way my 6 to 8 person team actually operates.
That’s the point.
AI for Loan Servicing
I’ve used The Mortgage Office as my loan servicing software for many years. It does what it needs to do. But the most recent upgrade I made is connecting it to Claude through an MCP server, built by my developer using The Mortgage Office’s public API.
Again, this doesn’t replace the software. It’s a faster access layer on top of it. Instead of logging into TMO to look up a loan balance or calculate a payoff figure, I can just ask Claude: “What’s the loan balance on 123 Main Street?” and get the answer immediately.
For quick lookups, that saves MEANINGFUL time across the day.
The more interesting use case is cross-referencing data. I can ask things like: “How many loans paid off in the last 8 months?” and get a clean answer without running multiple reports. Right now it’s more of a shortcut tool than a necessity, but the direction it’s pointing is exciting. fix-and-flip profits at 2008 lows, which means every efficiency gain matters more than it used to.
What I’m Trying to Build Toward
What I’d ultimately like to do is have Claude generate and send payoff statements directly. So instead of my team logging in, pulling the data, calculating the payoff, formatting the document, and emailing it, the whole thing happens in one command: “Email Chris the payoff statement for 123 Main Street.” Done.
That’s not fully built yet. But the infrastructure for it is largely in place. The data is in the servicing software.
The API connection exists. It’s a matter of wiring the output side. I imagine some of the newer servicing platforms already have something like this built in.
For lenders who are thinking through their long-term operational skill set, this kind of workflow automation is worth learning. The Federal Reserve is actively monitoring AI adoption across the financial sector, and the trajectory is clear.
McKinsey warns that just bolting new AI on top of existing processes without rethinking the workflow leads to technical debt, not transformation. That matches exactly what I’ve experienced. The MCP connections work because I built them around how my team already works, not around how some demo said I should work.
The Right Mindset for Implementing AI
Here’s the thing. If you go deep on AI tools, you will find an endless rabbit hole. New tools, new integrations, new capabilities, every single day.
I know people who have 8 AI agents running simultaneously and aren’t sleeping because they’re addicted to building more. That’s not what this is about.
My advice is simple: do one project at a time. Pick the thing that’s going to make the most meaningful difference in your operation, build it, tweak it until it works, and then add the next thing. Don’t try to implement voice AI, conversation AI, a custom origination platform, and MCP server connections all in the same month.
You’ll build nothing that actually works. That’s not theory. I’ve watched it happen.
For me, the projects that made the cut were the ones with a clear connection to revenue or cost reduction. More leads captured from missed calls. More leads captured from website visitors who don’t fill out a form.
Better reporting to know where to allocate marketing spend. Faster data access for underwriting calls. Those are concrete wins.
Most of the other ideas I’ve had got back-burnered because they were interesting but didn’t move the bottom line.
And here’s the other thing worth saying. A lot of people in the marketing and software world get nervous when they see AI tools becoming accessible to everyone. Like, now anyone can build their own software.
Yeah, but most people won’t. I built my origination platform because no existing software worked the way my team was already working. Not because I wanted to be in the software business.
The tool is so customized to my process that it would be almost useless to anyone else.
The edge isn’t in having the tools. It’s in actually knowing your operation well enough to deploy them in a way that fits. That knowledge comes from building a real lending business, not from chasing AI demos.
If you’re focused on scaling a real loan portfolio, that’s where your attention should stay. AI tools are in service of that goal, not the goal itself. Keep that straight.
It’s also worth noting that fix-and-flip profits at 2008 lows. That margin pressure on your borrowers is a reason for you, as a lender, to get more efficient. If you can process more deals with the same team, your unit economics improve even when origination volume stays flat.
If you want to get into the room where lenders are sharing what’s actually working, including the AI tools, the capital structures, the underwriting frameworks, come check out Hard Money Mastermind. I run monthly live coaching, a 2,600+ member network, and the full 15-hour Hard Money Masterclass covering everything from loan docs to raising capital. Real lenders, real deals, real systems.
Frequently Asked Questions
What AI tools are most useful for a hard money lending business right now?
The highest-impact tools I’m using are voice AI for missed inbound calls (converting about 80% to booked appointments), conversation AI chatbots on the website (roughly 3 new leads per day in early testing), large language models for content creation with human refinement, and MCP server connections to loan servicing and property data platforms like The Mortgage Office and SFR Analytics for faster data access.
What is an MCP server and how does it help a private lender?
An MCP server is essentially a connection layer that lets an AI assistant like Claude access data from another software through its API. For me, it means I can ask Claude plain-language questions about loan balances, transaction history, or borrower records instead of logging into multiple systems and running reports manually. It doesn’t replace the underlying software.
It makes the data faster to access and easier to cross-reference.
Should I use AI to generate content for my lending website?
Yes, but carefully. Use AI to identify the most relevant topics and create a content skeleton. Then have a human with actual lending experience refine, develop, and personalize that content.
Pasting raw AI output directly onto your site without human editing is likely to hurt your search rankings. Search engines are increasingly able to detect unedited AI-generated content.
How do I avoid getting overwhelmed when implementing AI tools in my lending business?
Do one project at a time. Pick the thing with the clearest connection to more revenue or lower costs, build it, tweak it until it works well, and then move to the next project. Trying to implement voice AI, a chatbot, a custom origination platform, and new servicing integrations simultaneously usually means nothing gets built properly.
Start small and add incrementally.
Do I need a developer to set up these AI tools or can I do it myself?
Some tools like GoHighLevel’s voice AI and conversation AI are built for non-technical users and can be set up without a developer once you understand the platform. The MCP server connections to platforms like The Mortgage Office required a developer working with the software’s public API. The custom origination platform also required development work.
If you’re not technical, budget for occasional developer help, especially for integrations between systems.
Please note that opinions stated by all speakers on this show are only opinions and not legal, financial, or accounting advice. Always check with your own legal, financial, and accounting professionals before making decisions for your business.
