A wave of AI-powered platforms has entered the real estate space. Some might be game changers, while others might be AI fluff.
This article is for real estate CEOs trying to understand where AI is creating real value and where it still struggles, the evolution of human-AI collaboration, and the key trends shaping what comes next.
This article draws from our white paper, Real Estate AI: A CEO’s Guide to What Matters Now. It tackles the most important AI questions facing real estate leaders and lays out a clear path to adoption. Download the full guide.
Four Core Capabilities Behind AI Platforms
Most AI software today are built around these four foundational functions:
Retrieve and Understand: AI systems excel at processing massive amounts of unstructured documents (contracts, reports, due diligence records, etc.) and extracting specific, organized information from them. This turns “document soup” into actionable insight.
Generate: AI produces new material based on provided context. This includes written content like property descriptions, technical outputs like spreadsheet formulas and database queries, and visual content like images and videos.
Predict: AI uses past information to make educated guesses about what’s likely to happen in the future. This includes predicting tenant churn, market pricing trends, and loan risk assessment.
Act: Advanced AI systems perform actions beyond generating outputs. This includes updating databases, creating and running code, or scheduling follow-up activities. These systems initiate changes rather than just providing analysis or predictions.
The Rise of Agentic AI
Traditional AI responds to specific prompts, but agentic AI takes action. The latter merges all four capabilities (retrieval, generation, prediction, and action) to plan and execute complex, multi-step processes. They don’t need to be told what to do, they figure it out themselves.
Consider agentic AI as a capable junior team member. It defines the problem scope, writes the query, retrieves relevant data, writes comprehensive reports, and handles follow-up communications - all without constant supervision.
While this autonomy level is emerging (refer to the below section), it shows where the technology is heading. In fact, it is already powering some of today's most effective AI applications.
SurfaceAI is one of them. SurfaceAI’s agents help multifamily teams uncover revenue, automate workflows, and run smarter operations.
The Evolution of Human-AI Collaboration
As these systems become more integrated into business operations, the relationship between people and AI typically develops through three phases:
Human in the Loop (HITL)
AI provides recommendations while humans make final decisions.
This approach works well when consequences are major, situations require complex judgment, or confidence in the system is still building. For example, an analyst reviewing AI-generated risk assessments before approval.
Human on the Loop (HOTL)
AI works on its own, but humans still guide and review its work. This setup allows teams to move faster while keeping human judgment in the loop.
AI can, for instance, generate a full investment memo and pass it along to the investment committee, with partners reviewing and signing off before it goes out.
Human Out of the Loop (HOOTL)
AI functions completely independently. Once properly configured, it manages everything (data gathering, decision-making, output creation, and follow-up actions) without requiring real-time human involvement.
For real estate companies, this means systems capable of handling tenant onboarding, deal evaluation, or pricing adjustments autonomously. While powerful, this requires robust safeguards, testing protocols, and governance frameworks.
This evolution isn’t universal. Some workflows, like legal reviews or high-value decisions, will likely always need human input. But by understanding the different ways AI can work, CEOs can better decide where AI should support human work versus where it can function independently.
Where AI Delivers Strong Results
Today's AI platforms consistently perform well in several key areas:
Summarizing Information and Generating Content
AI is great at breaking down complex or lengthy documents, like leases, internal reports, or market data, into clear, digestible summaries. It can also draft high-quality content, similar to what you'd expect from a junior associate: property listings, investor updates, job descriptions, executive overviews, and more.
When the format is familiar and repeatable, AI works quickly, adapts well, and delivers strong results.
Making Predictions From Clean, Structured Data
When the data is organized (think rent rolls, service records, or historical lease renewals) AI can spot patterns, flag outliers, and forecast likely outcomes. These aren’t just rough guesses. AI provides probability-based insights that help teams assess risk, plan outreach, and allocate time or budget more strategically.
Automating Repetitive, Rules-Based Tasks
Real estate workflows often involve tedious work; copying data, completing forms, tagging files, or replying to routine questions.
AI handles these tasks at scale, with more consistency than humans and zero fatigue. In many cases, it cuts manual effort by up to 80%, freeing up teams to focus on higher-value work.
Finding Answers Fast With AI Search
AI can search across contracts, emails, PDFs, and internal systems using natural language. Ask a plain question like, “What’s the lease renewal window for Unit 3A?” and it finds the exact line in the original document.
What AI Can’t Do Well (Yet)
While AI is powerful, it’s not suited for every situation, especially those that depend on human insight or real-world context. Here’s where caution is needed:
Handling Nuance and Human Judgment
AI can mimic reasoning, but it doesn’t understand emotion, relationships, or politics. It can’t build trust, negotiate complex deals, or navigate sensitive conversations. In any situation that depends on emotional intelligence or leadership instincts, people remain essential.
Dealing With Edge Cases and Unique Situations
AI performs best when the task follows a known pattern. But real estate is full of exceptions like custom legal clauses and one-off deal structures. These unpredictable scenarios confuse even the best models. When there’s no playbook, human oversight is still required.
Performing Exact Calculations and Logical Reasoning
Most AI models struggle with mathematics by design. They're built for language prediction, not calculation. This creates errors in multi-step logic, sequential calculations, or complex financial modeling.
However, this limitation can be addressed by connecting AI to specialized calculation tools and having AI trigger appropriate processes when needed. Leading systems use exactly this approach, surrounding the core AI with supporting infrastructure to ensure accuracy.
What to Expect Next from AI
AI is advancing fast, but not evenly. Some areas are seeing rapid improvement, while others are still hitting roadblocks.
As the tech matures, we’re seeing a shift in how AI fits into workflows. In some cases, it’s moving from humans actively guiding the process (human-in-the-loop) to humans just overseeing it (human-on-the-loop). And in select situations, it’s approaching full autonomy. But this shift isn’t happening all at once and not all tasks are ready to be handed off entirely.
Areas Likely to Improve Soon
Better grounding in real data: AI is becoming more reliable at pulling from actual data sources, reducing the chances of hallucinated answers.
Handling multi-step workflows: Newer agent-style systems are getting better at carrying out task sequences with fewer errors and more consistency.
Faster, more cost-effective performance: Infrastructure improvements are making AI cheaper and quicker to scale across teams and use cases.
Domain-specific expertise: Models trained specifically on real estate documents (leases, comps, valuations, etc.) are becoming more useful out of the box, with less fine-tuning needed.
Areas That Will Stay Limited (For Now)
Exact math and logic: AI still struggles with spreadsheet-level precision, and that’s unlikely to be solved immediately.
Understanding people and emotions: AI can’t sense tone, read between the lines, or interpret intent, so it falls short in situations that call for empathy or intuition.
Fully independent decision-making: High-stakes calls still require human judgment. AI can support, but accountability still sits with people.
De-Risking Your Next Move
Understanding where AI delivers value and where it falls short is just the beginning. As CEOs, the real challenge is pinpointing where the right opportunities exist in your real estate company and implementing solutions to solve them.
There are three ways to approach this:
Building in-house: A long, expensive road that demands hiring AI specialists or heavily investing in upskilling your current team.
Investing in a third-party AI platform: It all depends on the software you choose. There’s a big difference between legacy software with AI added on and platforms that are built around AI from the ground up.
Co-design a product you will love with Stackpoint: We partner with real estate companies like yours to turn your operational challenges into products that solve real problems and grow them into venture-scale companies.
The options are clear. If you’re looking to truly solve your pain points and co-build a solution that’s designed to scale, connect with us. We have done this countless times, ensuring fast insights, airtight data security, and minimal disruption.
