At Baseline, we’ve seen private lenders go from running their entire business on excel to now enthusiastically talking about AI. The shift seems to have happened overnight. As a technology company, this is very exciting for us to witness. Subsequently, one of the major items that has made its way onto our product roadmap is building vertical AI agents: AI systems tailor-made for specific industries. In our case, that industry is private real estate lending, often called “hard money”, “business-purpose”, or “non-bank” lending. These agents will transform how private lenders operate, compete, and serve clients.
Private real estate lending is a niche industry with nuanced workflows. Lenders place more weight on the collateral and the borrower’s ability to execute and manage a real estate investment rather than their tax returns or W2 income. Until recently, many private lenders cobbled numerous tools, often repurposed, to manage their workflows. But we began changing that with Baseline. We created a purpose-built solution that brought together over a dozen tools into a single platform. Now with vertical AI agents integrated into our purpose-built software, we have an opportunity to automate and enhance large swaths of the loan origination and servicing process.
So what exactly are vertical AI agents, and why should private lenders care? In this article, we’ll explore how they apply to private real estate lending. We’ll look at current applications bringing drastic improvements to the loan origination and servicing process, peek into the near future of what’s coming, and discuss how lenders can prepare their teams for an AI-powered world. Along the way, we will highlight examples, from both within and outside real estate, of companies building these solutions. Finally, we’ll share Baseline’s vision for vertical AI in private lending and what makes our approach unique.
What Are Vertical AI Agents?
A vertical AI agent is an AI system specialized for a particular industry or function. Unlike a general AI model (think of ChatGPT answering any question), a vertical AI agent is more like a digital expert in a specified field. It’s trained on domain-specific data, understands the industry jargon, and can perform tasks or make decisions within that context.
For example, a general AI tool might explain what a business-purpose loan is, but a vertical AI agent could actually help underwrite a business-purpose loan. It can evaluate the property risk, the borrower’s ability to successfully “rehab” the property, and ensure the terms you provide are in line with the state’s usury laws. The vertical agent has depth in its knowledge area. It doesn’t just chat; it actively works within workflows, handling specialized tasks end-to-end. In other words, these agents go beyond Q&A to get work done in the business.
Vertical AI agents leverage advanced machine learning, including LLMs and predicative analytics, tailored to lending requirements. They can handle processes like credit risk assessment, fraud detection, customer onboarding, loan servicing, and more, with exceptional precision. By focusing on a well-defined set of problems, such as evaluating a borrower’s rehab budget for a property, a vertical AI agent can perform better than a broad AI tool that lacks domain context. It understands the terminology (e.g. Blended LTC, ARV, escrow), the typical workflow steps, and the regulatory constraints of lending in a particular market.
From a broader perspective, we’re witnessing a shift in software from being just systems of record to becoming systems of action powered by these agents. In traditional software, humans input data and initiate actions. In the new model, the software, via AI agents, can take initiative by pulling data, analyzing it, and executing tasks, with some human supervision or even autonomously. We believe our vertical specialization will differentiate Baseline from other general purpose software tools. For private lenders, this means AI that truly understands their business.
Application of Vertical AI in Private Real Estate Lending Today
How can a vertical AI agent actually help a private real estate lender day to day? Let’s break it into two core parts of the business: loan origination (sourcing, processing, underwriting, and funding new loans) and loan servicing (managing loans through repayment). In both areas, early applications of AI are already delivering value and hinting at bigger changes to come
AI in Loan Origination
Origination involves many labor-intensive steps that are ripe for AI-driven automation and decision support. Here are some current real-world applications making a big difference:
Application Intake and Customer Engagement
A lot of data collection in private lending still happens via email, PDFs, or phone calls. AI agents are starting to streamline this. For instance, startups like Casca have introduced virtual loan officer assistants that handle the initial borrower interactions. Casca’s AI assistant “Sarah” converses with loan applicants via email or chat to gather all required information and documents to process a file. The AI agent works around the clock and keeps the process moving along. Any experienced lender understands the pain that comes with gathering information from busy borrowers which can take weeks of back and forth
Document Analysis and Data Extraction
Every loan file comes with a stack of documents like bank statements, valuations, insurance binders, title searches. AI equipped with natural language processing and computer vision can read and extract key data from these documents automatically, saving human processors countless hours. Lenders using Baseline use its document management functionality as a digital underwriting assistant, by automating the collection of documents and analyzing the highlights of key data points in a document as opposed to reviewing the document in its entirety. Another example is Blooma, an AI platform for analyzing commercial real estate, which ingests documents and pulls out relevant details to calculate relevant metrics. This means underwriters can get a comprehensive analysis and risk score of a prospective deal without spending days manually crunching numbers. The human underwriter can then focus on the higher-level judgement calls, with the grunt work handled by the AI.
Credit Risk Assessment
In private lending, credit risk is often less about the borrower’s FICO score and more about the asset and the borrower’s track record. Still, assessing risk is crucial. Vertical AI agents excel at analyzing large datasets to spot patterns that correlate with creditworthiness or default risk. Zest AI, for instance, is known for helping lenders incorporate machine learning in credit decisions to improve risk modeling and even reduce bias. In private lending, an AI agent can weigh property data, local market trends, the borrower’s experience (number of past successful flips, etc.), and even alternative data (e.g. borrower’s online presence) to produce a more nuanced risk profile than any single underwriting formula could. This assists in lenders making more informed go/no-go decisions and pricing the loan appropriately for the risk.
Property Valuation and Market Analysis
Private lenders live and die by collateral value. For some time now, we’ve had Automated Valuation Models (AVMs) that can instantly estimate a property’s value by analyzing comparable sales, local market trends, and satellite imagery. Proptech firms like RicherValues and HouseCanary offer such models to quickly get to a reliable valuation. RicherValues reports can predict the after-repair value (ARV) of a property based on the rehab budget provided by the borrower. AI can forecast future property values with greater accuracy than before, by detecting patterns in neighborhood-level data. This means a private lender originating an RTL product (often referred to as a fix-and-flip loan) can not only get today’s value, but also a prediction of the ARV once renovations are done, or how the market might move in the next 12 months. With that insight, a lender can adjust loan-to-value (LTV) ratios and terms more confidently. In short, AI offers information edge, seeing both the current landscape and the road ahead, which is incredibly powerful for loan origination decisions.
Fraud Detection
Fraud is an ever-present risk in lending. Fraudulent borrowers provide fake documents or contracts with inflated purchase prices. Here, vertical AI agents shine by catching subtleties humans often miss. They continuously learn from vast data points and can flag anomalies in loan applications. For example, an AI might notice if a bank statement submitted by a borrower has inconsistencies or if multiple loan applications with different names use the same IP address (potential identity fraud). AI-driven fraud detection runs in the background in real-time, flagging irregularities in documents or data that could indicate fraud.
Bringing all of the above together, AI agents dramatically compress the origination timeline. When you automate data entry and document review and get instant risk analysis, you can close loans much faster.
Today’s AI applications in loan origination are co-pilots to the human professionals. They automate grunt work, enhance analysis with deeper insights, and flag risks. Many of these capabilities are no longer experimental. They’re available in products like Baseline now and will only continue to improve.
AI in Loan Servicing and Portfolio Management
Servicing loans — tracking payments, providing statements, issuing payoffs, handling draw disbursements, and mitigating defaults — is another area where vertical AI agents are proving invaluable. Here’s how AI is changing loan servicing operations.
24/7 Borrower Support
Servicing often involves answering borrower questions. AI chatbots and voice assistants are stepping up here. Available through a portal or the lender’s website, these agents can provide immediate answers by pulling data from the system. This round-the-clock customer service improves borrower satisfaction and takes pressure off servicing teams. Companies like Intercom have successfully rolled out AI agents to provide automated customer support for software providers, where they handle many routine inquiries until the matter needs human attention.
Intelligent Default Prevention
Perhaps one of the most impactful uses of AI in servicing is early risk detection. AI can analyze data points and patterns to predict which loans are at risk of delinquency or default. For instance, it might flag that a borrower’s payment has been progressively later each month or that a dip in home prices or rental rates in a local market could affect the borrower’s ability to refinance or sell. These alerts allow lenders to proactively reach out and work with borrowers, like offering a loan modification or a refinancing option, before they default.
Construction or Rehab Draws
In private lending, lenders frequently provide loans with a rehab or construction draw component, where borrowers request draw releases as work on the property is completed. An AI agent could intake a draw request, automatically check it against the budget and schedule, may even analyze uploaded photos or inspection reports for work done, and either approve it or flag discrepancies for review. That level of automation can cut down a traditionally days-long process of phone tag and emails into a potentially near-instant decision. While we may be in the early days of such advance draw automation, the pieces (image recognition, anomaly detection) are already in place.
Regulatory Compliance and Reporting
Servicing requires careful compliance (e.g., sending correct notices, handling escrow properly, complying with state lending laws). Vertical AI agents can be programmed with these rules and can monitor the system to ensure compliance steps are not missed. They can also generate custom reports with natural language prompts. For example, if a lender needs to see that all loans have current hazard insurance, an AI can quickly scan the portfolio, verify insurance documents are on file for each loan, and compile custom notices to send to each borrower if any of them require attention. This would be a tedious project for a human team, but AI does it in seconds, drastically reducing risk and effort.
AI automates the mundane tasks, while keeping a vigilant eye on a lender’s portfolio, and enabling a high-touch experience without requiring a large staff. Many private lenders historically outsourced servicing because it was too cumbersome to manage in-house. With AI, even a lean team can handle servicing efficiently, maintaining control and visibility over their loans.
Near-Term Future: From Co-Pilots to Autonomous Agents
While current AI agents act mostly as co-pilots, where humans still oversee and finalize decisions, we are rapidly moving toward a future where these agents take on more autonomous roles. In conversations with lenders, we’ve observed that they’re open to using fully autonomous AI agents once trust is established, similar to how managers gradually provide autonomy to human employees after consistent performance and reliability.
Multi-modal AI Agents
So far, much of AI has been text or data-focused. The next wave is multi-modal agents that can process images, text, and voice together. For private lenders, this means an AI that can review an appraisal in PDF form, look at property photos, listen to a voice note from a borrower, and combine all that insight into actions. For example, after a site inspection, a borrower might send photos of a renovated kitchen along with their commentary. The multi-modal AI would then compare against the loan’s scope of work and recommend whether to accept or reject the draw request.
Predictive Portfolio Management
We touched on predictive analytics. Going forward AI will not just flag risks but also opportunities. An AI agent might identify that a group of short-term bridge loans in your portfolio could be refinanced into a DSCR loan or that certain geographies are seeing such price appreciation that it’s safe to offer slightly higher LTVs on new loans in those areas. Essentially, the AI will act as a portfolio strategist, constantly analyzing internal and external data to guide business strategy. Lenders who leverage this will gain competitive edge in pricing and customer retention.
Every Lender Get an AI Co-Pilot
Broadly, we foresee every role in a lending organization having an AI companion. Andreessen Horowitz, a leading venture firm, put it well: every white-collar worker will have an AI copilot, and some will have their roles fully automated by AI agents. In private lending, this could mean an AI for your loan officer, one for your processor, underwriter, servicing manager, and capital markets analyst, all specialized but working in concert. At Baseline we envision these agents integrated right into our platform so that hand-offs between stages are instantaneous and nothing falls through the cracks.
As confidence in AI agents grows, we’ll see more autonomy given to them. However, this comes with the need for explainability and oversight. Companies like ours are working to ensure AI recommendations come with rationale (e.g., “Loan not approved because DSCR was above threshold and rents in area are declining”). This fosters trust so that eventually, fully autonomous decisions, like auto-declining or approving loans within certain parameters can happen without trepidation.
Preparing Your Lending Business for AI Adoption
Adopting AI agents in current private lending operations will likely require a shift in strategy in terms of team structure, skills, and workflow integrations. Based on our experience working with private lenders, here are some practical steps to position your business.
Invest in Data Infrastructure
AI is only as good as the data it has. Take a hard look at how you capture and store data today. Are your borrower records, investor reporting, and loan performance data in digital, structured formats accessible to AI tools? If you’re still reliant on spreadsheets, consider moving onto a modern loan management platform like Baseline that centralizes your data. Consolidating your operations onto one system not only improves workflow efficiency, it lays the groundwork for AI because all the relevant data is in one place. The ideal scenario is an AI agent living in your system of record, where it can easily fetch and update information. So, focus on data cleanliness and digitization, because it’s critical for an AI agent to be deployed successfully.
Start with High-Impact, Manageable Projects
You don’t have to implement AI in every part of your business overnight. Identify a few pain points in your workflow that are relatively straightforward for AI to tackle. For example, many private lenders start with intelligent document processing, using AI to scan documents and providing a summary of key data points. Another common starting point is a chatbot on your website to answer borrower FAQs or collect initial deal info. These types of projects have clear ROI, as staff saves time on low-value tasks, and are less risky. As you gain confidence, you can expand AI’s role into more sensitive areas like credit decisioning or portfolio analytics.
Upskill and Hire for the Future
Your team structure might need to evolve. Consider appointing an “AI Lead”, that is someone who will be the point person for understanding AI tools and working with vendors. This doesn’t have to be a new hire. It could be a tech-savvy operations manager who is given the mandate and time to focus on AI projects. Larger organizations might even hire an AI specialist, but for many, partnering with technology providers and having an internal champion coordinating is sufficient. When hiring new team members, consider a blend of domain expertise and tech savvy. The lender of the future might have an underwriter who also knows how to write a simple prompt or query to an AI. In short, add digital skills as a requirement alongside the traditional lending skills.
Integrate AI into Your Workflow with Human Oversight
As you deploy AI agents, integrate them in a way that fits your operation. This often means keeping humans in the loop initially. For instance, you might have the AI auto-generate loan recommendations, but a credit committee still reviews and approves them. Or an AI chatbot might handle borrower interactions up to a point, but routes the conversation to a human if it gets complex. This not only safeguards your processes but also builds trust in the AI’s reliability. Over time, as confidence grows, you can dial back the level of manual oversight and let the AI agent assume more independence.
Partner with the Right Technology Provider
Building a vertical AI agent from scratch in-house is generally not feasible for most private lending firms. There’s a growing ecosystem of vertical AI solutions and vendors building powerful AI tools. Evaluate partners that understand your business and industry. When evaluating, key things to look for include: does the AI integrate with your existing software? How does it handle data security and privacy? Does the provider offer support to help your team learn and refine the system? Given the nascent state of AI in our industry, a strong partnership will go a long way. As Baseline, for example, we work closely with lenders when introducing new AI features, incorporating their feedback to fine-tune the agent’s behavior. You want a provider that will effectively co-create value with you, not just sell and disappear.
For private lenders, the goal should be to be proactive. Start laying the groundwork now, so that as advanced AI capabilities become available, you can plug them in quickly. The learning curve is real, but manageable, and the upside is substantial for those who prepare.
Baseline’s Vision
Our mission has always been to empower private real estate lenders with technology that levels the playing field. We’ve seen how our software helps lenders deliver a digital experience on par with the big banks. In many ways, the technology we offer has far exceeded the experience users get with archaic banking software. Now, we’re taking the next step by infusing vertical AI deeply into that software.
Baseline is building AI agents that are native to our platform, meaning they live within the same system private lenders use for processing applications, underwriting, servicing, investor management, and so on. This is crucial. Because our AI has direct access to the lender’s data, with all the necessary permissions and security, it can act with full context. When our AI co-pilot assists in underwriting, it doesn’t operate on a generic dataset. It looks at your historical loan outcomes, your specific credit guidelines, and even your region’s market data. It’s like hiring an analyst who has worked in your company for years and knows exactly how you do business. This yields more relevant recommendations than any off-the-shelf AI could. Baseline’s vertical AI agent isn’t a one-size chatbot. It’s an embedded expert tuned to the private lending domain and to each lender’s nuances.
While most point solutions tackle one slice of the workflow, Baseline’s vision is an integrated AI-driven workflow from start to finish. Since our platform already handles everything from loan origination to servicing, our AI agents can seamlessly hand off tasks along the value chain. Picture our AI agent screens an application and pre-qualifies the borrower, then automatically orders an appraisal or pulls an AVM for the collateral, then assists the underwriter in structuring the deal, generates the term sheet, and after closing, continues to monitor the loan in servicing. All of this happens in one system. This continuity is a big differentiator, which means no lost information and no need for different AI tools at different stages, and a consistent experience for the lender and borrower alike.
At the end of the day, private lending is a relationship business. Our approach to AI is to enhance the human relationship, not replace it. By offloading administrative tasks to an AI agent, Baseline’s tools let lenders spend more time on nurturing client relationships, raising capital, and making strategic decisions. Lenders scale their business by developing stronger relationships with brokers and borrowers, not in front of a spreadsheet.
Baseline’s approach is unique because we live and breathe this domain. We’re not a generic AI toolkit trying to learn private lending from scratch. We’re a private lending platform infusing AI into the DNA of a system that already understands your business. We believe this will deliver superior results.
For years, the private lending industry lagged behind other industries in tech adoption, and relied on manual processes and fragmented software. Vertical AI agents offer a chance to leapfrog from spreadsheets or old software to tools built for the next generation. The lenders who embrace AI and integrate it into their workflows will likely see higher profitability and scalability than those who maintain status quo, simply because they can handle more business with fewer errors and faster turnaround.
