Artificial intelligence (AI) has moved fast in commercial real estate (CRE). In just a short period of time, tools like ChatGPT and other large language models have introduced new ways to analyze information, generate content, and accelerate everyday work. For many firms, this initial exposure sparked excitement — and just as quickly, hesitation. The question for most firms isn’t whether AI works. It’s about whether it can be trusted to handle the financials, underwriting assumptions, tenant data, and investment strategies. As a result, many CRE firms have found themselves stuck between two extremes: broad experimentation with general-purpose AI tools or a complete pause on AI adoption altogether.
The firms gaining real ground aren’t chasing hype or banning AI entirely, they’re investing in true AI adoption and readiness: governed data, controlled workflows, and purpose-built intelligence for CRE operations.
Want to listen to the discussion? Click here to watch the on-demand webinar.
The First Year of AI in CRE: What Worked and What Didn’t
Most early experimentation centered on general-purpose foundational models, useful for surface tasks but never engineered for institutional-grade CRE work. True AI adoption requires moving beyond experimentation to integrate these tools into real workflows. These are general-purpose tools trained on vast amounts of public internet data. They excel at tasks like writing emails, summarizing documents, brainstorming ideas, or performing high-level research.
For individual productivity, they proved valuable. Analysts used them to prepare for meetings. Asset managers used them to draft reports. Executives used them to explore market trends or sharpen talking points. In that sense, AI demonstrated its potential to multiply time, raise baseline competency, and help professionals walk into meetings better prepared.
However, cracks began to show as firms pushed these tools deeper into their workflows.
Foundational models weren’t built for institutional CRE workflows — not without private hosting, integration, and governance layered on top. They do not natively integrate with property management systems, accounting platforms, debt models, or investor reporting tools. They require manual data uploads, which quickly become burdensome at portfolio scale. More importantly, they introduce serious data governance and security risks when proprietary information is entered directly into public systems.
For organizations responsible for safeguarding tenant data, investor information, and strategic plans, these risks are not theoretical. When firms rely on public AI, they often lose visibility into how their data is processed or stored — a risk no owner or operator can accept. Even anonymized data can introduce compliance concerns. As a result, many large real estate firms have restricted or outright banned the use of public AI tools for business-critical work.
Understanding the Difference: Foundational AI vs. Applied AI
The distinction between foundational AI and applied AI is central to understanding where the industry is headed.
Foundational models are broad and unfocused by design. They answer general questions well but lack deep context about specific industries, datasets, and workflows. They operate at the top of the funnel, providing wide but shallow intelligence.
Applied AI moves from general intelligence to operational execution, combining domain expertise with a firm’s private data, governed workflows, and business logic. Effective AI adoption ensures these systems are integrated into CRE operations rather than treated as a separate tool. Instead of scraping the internet, applied AI analyzes live operational data, rent rolls, historical financials, budgets, underwriting models, and debt documents all within a secure environment.
The output is not just faster answers, but better ones. Applied AI understands the difference between net operating income (NOI) and cashflow, debt service coverage ratio (DSCR) and debt yield, Class A and Class B assets, floating-rate risk and forward curves. It is designed to operate inside the realities of commercial real estate rather than speaking about them abstractly.
Why Data Security Is the Non-Negotiable Requirement
Data security is the dividing line between experimentation and adoption.
Data volume is enormous and constantly changing across CRE portfolios and operations. Rent rolls update monthly. Financials roll forward. Leasing velocity shifts. Capital expenditures evolve. Debt terms, covenants, and hedging structures introduce another layer of complexity. Manually uploading this information into a general AI tool not only erodes efficiency — it undermines the value proposition of AI itself.
Applied AI platforms solve this by operating inside a firm’s private data environment, not pushing data into public systems. Information flows into a secure data warehouse through integrations with property management systems (PMS), accounting platforms, underwriting models, and document repositories. The AI model analyzes the data without absorbing it into a global training set. The data stays siloed, governed, and auditable.
This structure allows firms to use AI confidently, without worrying about exposing tenant records, loan documents, or investor communications to unintended audiences.
From Answers to Action: The Role of AI Agents
One of the most powerful evolutions within applied AI is the rise of AI agents.
An AI agent functions like a role-based digital team member — an analyst, an asset manager, or a capital markets expert with defined responsibilities. Instead of a generic chatbot, an agent operates with specific instructions, domain expertise, and objectives. In CRE, this might include an accounting agent reviewing trial balances, an asset management agent evaluating operational performance, or a capital markets agent stress-testing a debt scenario.
Agents combine role, task, and data. They don’t simply answer questions; they execute analysis within defined guardrails. This allows AI to mirror the structure of a CRE firm, where different teams focus on different responsibilities but rely on shared information.
As agents become more specialized, the return on investment increases. AI moves from being a novelty to becoming a multiplier across underwriting, refinancing, asset management, and investor relations.
Prompting Still Matters — But the Burden Drops
Prompting gets framed as a technical skill, but in practice it’s management — directing AI the same way you guide a team member.
Clear context, defined roles, specific instructions, and desired outcomes lead to better results. When AI is treated like a junior team member (guided, corrected, and refined through conversation), the quality of output improves quickly.
Applied AI reduces the cognitive load of prompting because it already understands the domain and the data. Users no longer need to explain what a rent roll is or why delinquency matters. Over time, the system adapts to user preferences, reporting styles, and analytical priorities.
By providing consistent instructions and clear context, teams can generate repeatable, high-quality outputs that align with organizational standards. This approach ensures that reporting, analysis, and operational insights are reliable, actionable, and scalable across the portfolio.
Download the free checklist: Query Your Way to Success – A Step-by-Step Guide to Leveraging AI in CRE
Real Efficiency Gains at Portfolio Scale
The true power of applied AI becomes clear at scale.
Rather than manually assembling twelve months of financials across ten properties, users can query an entire portfolio instantly. Instead of preparing investor letters from scratch, AI can draft performance narratives aligned with actual asset data. Instead of reactive asset management, teams can surface risks, trends, and opportunities proactively.
AI also enables workflows, not just one-off insights. Action items can be tracked. Follow-ups can be scheduled. Reporting expectations can be formalized. This transforms AI from a research tool into an operating system for decision-making.
The Path Forward for CRE Firms
Commercial real estate is moving from experimentation to execution. The firms that gain real advantage will be those building strong foundations: secure data, applied intelligence, role-based agents, and workflows designed for real decisions. The barrier to entry is lower than ever, but the strategic payoff of doing it right has never been higher.
AI is already reshaping how firms assess risk, manage assets, and allocate capital. The question is no longer whether to adopt it, but whether to implement it responsibly, securely, and with intention. Firms that approach AI adoption strategically will define the next chapter of the industry.