You don't need to build the engine to manage the building
This module is not a computer science lecture. You will never need to train a model or write code.
What you need is a working mental model — an understanding of what AI can do with your lease documents, your appraisal reports, and your market data. And equally important, an understanding of where it will confidently give you wrong answers.
Think of AI the way you think about a new hire: capable, fast, and eager — but needing your direction, your context, and your review.
Which is the best analogy for how you should think about AI in your real estate practice?
What an LLM actually does — in real estate terms
A large language model is a pattern recognition system trained on enormous amounts of text. Imagine an analyst who has read every public lease, property management manual, appraisal textbook, zoning code, real estate legal opinion, and market research report ever published — and developed an extraordinarily refined sense of how real estate language works.
That is roughly what an LLM is. It does not "know" things the way you do from walking a property or negotiating a lease. It recognises patterns and generates responses that follow those patterns.
Excellent at:
- Synthesis — combining information from a 120-page lease, a rent roll, and local market data into a coherent summary
- Extraction — pulling specific terms from unstructured documents (rent, escalation schedules, renewal options, use restrictions)
- First drafts — well-structured lease abstracts, property descriptions, market reports, tenant communications
- Comparison — identifying differences across multiple leases, appraisals, or offering memos
Unreliable at:
- Current market data — training data has a cutoff date; it cannot tell you today's cap rates in your submarket
- Precise calculation — it is a language model, not a spreadsheet; always verify math in Excel
- Factual claims without source material — it can "hallucinate" zoning regulations or make up comparable sales
The context window — how many lease pages can AI read at once?
The context window is how much information AI can hold in working memory during a single conversation. Think of it as the stack of documents you can hand to your AI analyst before it starts working.
Current models handle approximately 200,000 tokens — roughly 500 pages in a single conversation.
| Context size | What it holds | Real estate use case |
|---|---|---|
| Small (~8K tokens) | 3-5 pages | Quick question about a single lease clause |
| Medium (~32K tokens) | 40-50 pages | Full residential lease review |
| Large (~200K tokens) | ~500 pages | Entire commercial lease with amendments, rent roll, and market report |
What this means for you: You can upload an entire 150-page commercial lease with all its amendments and ask AI to extract every rent escalation clause, renewal option, and exclusivity provision in one conversation. You do not need to break it into pieces.
Critical insight: The quality of AI output is directly proportional to the quality of the documents you provide. Give AI the actual lease — do not ask it to work from memory or summarise "typical" lease terms. It works best when it can read the source material.
Real estate capabilities — what works today
These are not theoretical. These are workflows that real estate professionals are using AI for right now.
Lease abstraction — Upload a commercial lease and get a structured summary of every material term: base rent, escalations, options to renew, options to terminate, use restrictions, assignment and subletting provisions, co-tenancy clauses, CAM caps, and more. What takes 2-4 hours manually takes 10-15 minutes with AI.
Comparable property analysis — Give AI a set of recent sale comps or lease comps and it will organise them into a structured comparison, identify outliers, and draft the comp analysis section of an appraisal or broker opinion of value.
Market report drafting — Provide AI with market data (vacancy rates, absorption, new supply, rent trends) and it will produce a polished market overview section for a listing presentation or investment memo.
Appraisal report review — Upload an appraisal and ask AI to extract the key value conclusions, flag any inconsistencies between the income and sales comparison approaches, and summarise the appraiser's assumptions.
Tenant communication drafting — Generate personalised lease renewal letters, maintenance updates, rent increase notifications, and other correspondence across a portfolio of tenants.
Which of these tasks would AI be LEAST reliable at performing without verification?
What AI cannot do — and what still requires you
Knowing the limits is as important as knowing the capabilities.
AI will not predict market crashes or timing. It can summarise historical market data and identify trends in the numbers you provide, but it cannot tell you when the next downturn will hit or whether a submarket will outperform. Market judgment is yours.
AI cannot replace site visits. It can analyse a property condition report, but it cannot smell mould, hear the traffic noise, notice the deferred maintenance the inspector missed, or feel the neighbourhood vibe that tells you whether a retail location will succeed.
AI does not know your local market the way you do. It may have general knowledge about a metro area, but it does not know that the block behind the property has three new developments in permitting, or that the municipality is about to rezone the corridor. Your local expertise is irreplaceable.
AI makes confident-sounding errors. This is the most dangerous limitation. AI will never say "I'm not sure." It will present incorrect information with the same confidence as correct information. A hallucinated zoning classification or an invented comparable sale looks identical to a real one in the output. You must verify.
Data privacy — protecting tenant PII and confidential deal information
Real estate data includes some of the most sensitive information in any industry — tenant personal information, financial statements, social security numbers from rental applications, bank account details, and confidential deal terms.
Enterprise AI plans (Claude for Enterprise, ChatGPT Enterprise, Microsoft Copilot for Enterprise):
- Your data is not used to train the model. Full stop.
- Encrypted in transit and at rest
- Access controls and audit logs
- Data residency options for regulated environments
The real risk: Someone on your property management team using a free personal AI account to process tenant applications containing social security numbers, pay stubs, and bank statements. Free-tier AI services may use your inputs for training and lack enterprise-grade security controls.
What you need to do:
- Use enterprise-tier AI tools for any workflow involving tenant PII, financial data, or confidential deal terms
- Establish a clear policy about what data can and cannot be processed through AI — and which tools are approved
- Redact when possible — if you only need AI to review lease terms, strip the tenant's personal financial information before uploading
- Audit shadow usage — assume that people on your team are already using free AI tools with sensitive data; provide a sanctioned alternative immediately
A leasing agent wants to use a free AI chatbot to summarise rental applications. What should you do?
Key takeaways
- AI is a fast junior analyst, not magic — it needs your direction and review.
- The context window lets you process entire commercial leases and due diligence packages in a single conversation.
- AI excels at extraction, synthesis, comparison, and first drafts of real estate documents.
- AI is unreliable for math, market predictions, site-specific observations, and current market data.
- Data privacy is solvable — use enterprise-tier tools and establish clear policies for tenant PII.
Next up: AI for Lease Management.
Module 2 — Final Assessment
01What is the best analogy for an LLM's context window in a real estate context?
02Which task should you NEVER trust an LLM to do without verification?
03What is the primary data privacy risk when using AI in property management?