I recently attended a panel at the California MBA Western states CREF Conference titled AI, Relationships & Commercial Mortgage Originators. (actually it was a couple of months ago but was swamped)
Daniel Ceniceros of Connect CRE hosted the panel, which featured Jason Berry of Slatt Capital, Andrew Pollock of Renovo Financial, and Sammy Greenwall of Henry AI.
Most conversations about artificial intelligence eventually drift toward some distant future in which machines do everything.
This conversation was grounded in what lenders are doing today.
AI is already reading rent rolls, extracting operating statements, preparing underwriting models, generating marketing packages, researching borrowers, updating Salesforce, conducting background checks, reviewing inspections, recording meetings, and producing reports.
The work is changing quickly.
The central question is what happens to the people.
The First Wave Is Already Here
Slatt Capital closes approximately one loan every business day across the middle market and small balance lending sectors. At that volume, efficiency becomes a meaningful competitive advantage.
Jason Berry described the immediate opportunity clearly.
“The challenge in our industry is that all the data is locked up either in Excel or emails.”
That data then has to be extracted, organized, checked, and entered into the different systems required by borrowers, intermediaries, servicers, and lenders.
A borrower sends a rent roll.
An analyst enters it into a sizing model.
The numbers move into a presentation.
A lender requests the same information through its portal.
An underwriter transfers it into another model.
An investment committee receives another version.
The same information passes through multiple hands and multiple systems.
Artificial intelligence can remove much of that repeated work.
Slatt has built an AI process that takes rent rolls and operating statements and loads the information into its loan-sizing model. An analyst reviews the output and applies judgment. The time once spent entering numbers can now be spent understanding them.
Berry explained why this matters:
“By far, the number one complaint everybody had was, ‘I don’t like doing the data entry.’ That’s a huge problem in our industry.”
The Most Interesting New Hire May Be an AI Analyst
Slatt Capital has also hired an AI business analyst.
His job is to work with analysts, producers, marketing, and servicing to identify repetitive work and develop practical AI applications.
That may become one of the most important new roles in commercial real estate finance.
Most employees already have full workloads. They cannot spend every Friday experimenting with Claude, OpenAI, Copilot, skills, and new connections into Salesforce. The technology changes too quickly and the number of possible applications is too large.
Someone needs to own the process.
The AI analyst studies how the firm works, identifies easy wins, tests solutions, and helps employees adopt them. The role sits between operations, technology, and the business itself.
This is less about buying software and more about redesigning work.
The Breakthrough Is the Connection
One of Slatt’s most significant recent improvements came through an MCP connection to Salesforce.
MCP — Model Context Protocol — allows AI systems such as Claude and Copilot to communicate directly with company software and data. Employees can ask questions in plain English rather than running reports and navigating databases.
Berry gave a simple example. A servicing executive who previously ran Salesforce reports can now ask the system a question directly. The AI queries the database and delivers the answer in Excel, PowerPoint, or another requested format.
“That has changed things dramatically for our firm in just the last month.”
This is where AI becomes much more powerful.
A standalone chatbot can help an individual complete a task. A connected AI system can understand the company’s data, processes, customers, and history.
Every lender has accumulated an enormous amount of institutional knowledge. Much of it remains trapped in email accounts, spreadsheets, recorded meetings, personal contact lists, and the minds of experienced employees.
Slatt took hundreds of recorded internal training meetings and used AI to turn them into a company training manual.
The knowledge already existed. AI made it accessible.
Clean Data Becomes a Competitive Advantage
AI depends on the quality of the information it can access.
Andrew Pollock described Renovo Financial’s focus on moving information electronically through its systems.
“When you get the human hand off the keyboard, the integrity of the data becomes less contaminated.”
Every manual transfer creates an opportunity for error. A number gets entered incorrectly. A field is left blank. A borrower is duplicated. A lender’s criteria become outdated. A document gets saved in someone’s inbox and disappears from the company’s collective memory.
AI can process thousands of records quickly. It can also distribute a bad piece of information across thousands of records quickly.
The firms that organize their data now will gain an enormous advantage.
Their systems will know which lenders finance a particular asset class, which sponsors perform, which markets are changing, which loans are approaching maturity, and which borrowers may need capital.
They will also be able to identify opportunities earlier.
Pollock described using entitlement and permitting data to find developers who may need financing long before they begin approaching lenders.
“Now we can see the opportunity to provide financing well in advance of when they need it.”
That is the future of origination.
The strongest originators will arrive earlier, know more, and provide value before asking for the business.
The Lender Database Needs a Pulse
The industry has spent years trying to build the perfect lender database.
Lender appetite constantly changes.
A lender may finance hotels while reaching its hotel allocation for the quarter. It may offer an attractive program and lack the capacity to underwrite another loan. It may be active nationally while pulling back from a particular market. It may quote aggressively while closing slowly.
Berry questioned the idea of a magical lender database because knowing a lender’s published criteria only tells part of the story.
“You need to know the lenders really well. Sometimes they’re full. Sometimes they’re active. Sometimes they don’t quote as quickly as possible.”
A static database captures what a lender says it does.
A living lending network captures what the lender is actually doing.
That requires recent quotes, response times, closed loans, changing allocations, geographic preferences, and direct conversations with the people making decisions.
AI can organize and interpret those signals. Relationships create the signals.
The winning lending platform will understand the property, borrower, transaction, lender, and timing. It will see demand forming and connect the right parties while the opportunity is still developing.
Analysts Will Learn a Better Business
Junior employees have traditionally learned commercial real estate finance by spreading rent rolls, entering expenses, formatting presentations, and moving information between systems.
The work exposed them to transactions, although much of their time was consumed by data entry.
AI creates a better apprenticeship.
Analysts can spend more time studying the property, questioning assumptions, speaking with borrowers, understanding lender preferences, evaluating risk, and participating in structuring decisions.
Sammy Greenwall sees this as a faster path from analyst to producer.
“The things they need to focus on have changed from spreadsheets to relationship building and truly understanding the fundamentals.”
The ability to build a spreadsheet will carry less value.
The ability to recognize when the spreadsheet is wrong will carry much more.
Experienced professionals possess context and judgment developed over decades. AI can produce an answer with extraordinary confidence while missing a fact that an experienced lender would recognize immediately.
Human review remains essential because confidence and accuracy are very different things.
Adoption Begins With an Obvious Win
Large AI strategies can stall because they require too much time, integration, and organizational change.
Adoption happens quickly when a tool solves a problem employees already dislike.
“You just saved me two hours on every analysis I have to do, and the output is so much better.”
That is the standard every AI investment should meet.
Does it solve a clear problem?
Does it fit into the existing workflow?
Can employees see the benefit immediately?
Can the company measure the result?
Small successes create confidence. Confidence creates adoption. Adoption creates the foundation for larger changes.
Build the Experiment. Buy the Infrastructure.
AI tools now make it possible for an individual to build a useful workflow in fifteen minutes.
That can create a dangerous illusion.
A process that handles one rent roll successfully may fail when asked to handle one thousand. Enterprise systems require testing, security, controls, permissions, audit trails, and dependable performance.
Berry drew a useful line between experimentation and scale.
“We can probably do 50 percent of the work product ourselves, but we’re going to need enterprise systems designed by engineers.”
Lending companies should develop the experiments that help them understand their needs. Proven vendors should provide the infrastructure required to run those processes safely at scale.
Trust Will Decide the Winners
Commercial real estate lenders hold sensitive borrower information, financial statements, ownership records, personal information, and proprietary credit decisions.
Employees are already placing some of that information into consumer AI accounts. Many have little understanding of where the information is stored or how it may be used.
Companies need secure enterprise accounts, approved tools, controlled access, clear policies, and vendors capable of explaining their architecture.
Greenwall emphasized the importance of asking vendors to prove how information is protected.
“Make sure that the vendor can prove the architecture they’ve built and the certifications they have.”
Data security will become part of the lending relationship.
Borrowers will want to know who can access their information. Lenders will want to understand how an AI system reached its conclusion. Regulators and investors will expect an audit trail.
Trust has always mattered in lending. AI expands its meaning.
The Most Human Lender May Win
Greenwall believes firms that embrace AI will generate significantly more revenue per employee.
That feels inevitable.
Lenders and intermediaries will identify opportunities earlier, process information faster, evaluate more transactions, and respond to borrowers with greater speed and precision.
The technology itself will eventually become widely available.
The lasting advantage will come from how people use the time it creates.
The best originators will spend more time meeting borrowers, understanding lenders, visiting properties, solving difficult structures, and staying close to their markets. The best analysts will develop judgment sooner. The best companies will turn their collective knowledge into a system that makes everyone more capable.
Daniel Ceniceros ended the panel with advice a friend in the AI industry had given him about preparing the next generation:
“Teach them how to be human. Get them off their phones. Teach them how to walk across the room and shake hands.”
Commercial real estate lending has always moved at the speed of information and trust.
AI is about to accelerate the information.
The opportunity is to invest the time it creates in trust.



