Claude Opus 4.5 just dropped. Here are 7 ways to use it in real estate today.
Anthropic released their most powerful model today. I translated it into use cases you can test this week.
Claude Opus 4.5 launched today.
If you’re like most people...
You saw the announcement.
Skimmed the benchmarks.
Thought “cool, another AI update.”
And moved on.
I get it.
The AI news cycle is exhausting.
But this one’s different.
And I’m going to show you exactly why — in terms you actually care about.
Here’s what Anthropic shipped:
According to their official announcement, Opus 4.5 delivers:
State-of-the-art Excel automation and financial modeling
20% accuracy improvement on complex spreadsheet tasks
15% efficiency gains on document creation
New Chrome browser integration
“Infinite chat” (no more context window limits)
Agents that can refine their own workflows
That’s what they said.
Here’s what it means for you.
7 Real Estate Use Cases You Can Test This Week
I went through Anthropic’s release and translated every capability into something you can actually use.
Credit where it’s due — these are based on Anthropic’s published use cases. I just made them relevant to your business.
1. Underwriting Models That Actually Work
What Anthropic said: “Claude Opus 4.5 sets a new standard for Excel automation and financial modeling.”
What this means for you:
You can now ask Claude to build or modify your underwriting models with real logic.
Not summaries.
Actual cell formulas.
Input checks.
Sensitivity tables.
Try this prompt:
You are a senior real estate acquisitions analyst.
Build me a multifamily underwriting model in Excel with:
- Purchase price, units, and average rent inputs
- Operating expense assumptions (taxes, insurance, management at 5%, repairs at 5%, vacancy at 5%)
- Debt service calculations (loan amount at 75% LTV, 7% rate, 30-year amortization)
- NOI, cash flow, cap rate, and cash-on-cash return outputs
- A sensitivity table showing cash-on-cash returns at 3 different purchase prices and 3 different interest rates
Use proper Excel cell references. Label all assumptions clearly. Make it easy for me to modify inputs.2. Chrome-Based Deal Research (Without Tab Chaos)
What Anthropic said: “Claude for Chrome lets Claude handle tasks across your browser tabs.”
What this means for you:
Claude can now read, extract, and organize information from multiple browser tabs.
LoopNet.
CoStar.
County assessor sites.
Redfin.
All at once.
Try this prompt:
I have 5 multifamily listings open in my browser tabs.
For each property, extract:
- Address
- Asking price
- Number of units
- Year built
- Price per unit
- Listed cap rate (if available)
- Any notable features or red flags mentioned
Organize this into a comparison table sorted by price per unit from lowest to highest.
Flag any properties missing key information.3. Investor Reports in Minutes
What Anthropic said: “Opus 4.5 can produce documents, spreadsheets, and presentations with consistency, professional polish, and domain awareness.”
What this means for you:
Monthly investor updates.
Quarterly portfolio summaries.
LP communications.
All generated from raw data you already have.
Try this prompt:
You are an asset manager preparing a monthly investor update.
Here’s my property’s data for October:
- Property: Oakwood Apartments, 24 units, Phoenix AZ
- Occupancy: 92% (22/24 units)
- Gross potential rent: $38,400
- Collected rent: $34,560 (90% collection rate)
- Operating expenses: $12,200 (budgeted $11,500)
- Maintenance: Replaced 2 HVAC units ($4,800 unbudgeted)
- Market note: Phoenix multifamily vacancy rose to 8.2% metro-wide
Create a one-page investor update that includes:
- Executive summary (2-3 sentences)
- Occupancy and rent collection summary
- Key variances from budget with explanations
- Maintenance and CapEx highlights
- Market context
- Outlook for next month
Tone: Professional but conversational. No jargon. Write like you’re updating a trusted partner.4. Lease Abstraction That Doesn’t Miss Details
What Anthropic said: “Opus 4.5 excels at complex, long-running workflows requiring sustained reasoning.”
What this means for you:
Upload a 40-page commercial lease.
Get a complete abstract.
Rent escalations.
CAM provisions.
Renewal options.
Termination clauses.
Co-tenancy requirements.
The model now maintains context across the entire document.
No more “I don’t see that section” errors.
Try this prompt:
You are a commercial real estate paralegal specializing in lease abstraction.
I’m uploading a commercial lease. Create a comprehensive lease abstract that includes:
TENANT & PREMISES
- Tenant legal name and trade name
- Premises address and suite number
- Square footage (rentable and usable if stated)
- Permitted use
TERM
- Lease commencement date
- Lease expiration date
- Any extension or renewal options (terms and notice requirements)
RENT
- Base rent schedule (all amounts and escalations)
- Percentage rent provisions (if applicable)
- CAM/NNN obligations and caps
- Security deposit amount
KEY PROVISIONS
- Assignment and subletting restrictions
- Exclusivity clauses
- Co-tenancy requirements
- Early termination rights
- Personal guaranty terms
- Insurance requirements
Flag any unusual provisions or missing information I should follow up on.Quick question:
Which of these would save you the most time this week?
Hit reply and tell me.
I read every response.
5. Due Diligence Checklists That Build Themselves
What Anthropic said: “Claude asks clarifying questions upfront, then builds a user-editable plan before executing.”
What this means for you:
Tell Claude the deal type.
It asks what you need to verify.
Then it generates a customized due diligence checklist.
With responsible parties.
Deadlines.
Document requirements.
Try this prompt:
You are a real estate acquisitions manager helping me build a due diligence checklist.
Deal details:
- Property type: 24-unit apartment building
- Location: Dallas, Texas
- Purchase price: $3.2M
- Closing timeline: 45 days
- Financing: Agency loan (Fannie Mae)
Before creating the checklist, ask me 3-5 clarifying questions about:
- The property’s age and condition
- Whether it’s occupied or vacant
- Any known issues disclosed by seller
- My team structure (who handles what)
Then create a comprehensive due diligence checklist organized by category:
- Title & Survey
- Physical / Property Condition
- Financial / Rent Roll
- Legal / Lease Review
- Environmental
- Insurance
- Lender Requirements
Include typical timeline estimates and responsible party assignments.6. Comparable Analysis From Messy Data
What Anthropic said: “Accuracy on internal evaluations improved 20%, efficiency rose 15%.”
What this means for you:
You can dump a messy CSV of recent sales.
Claude cleans it.
Standardizes the fields.
Calculates price per unit, price per square foot, cap rates.
And produces a formatted comp table.
This used to require multiple prompts and manual cleanup.
Now it handles the entire workflow.
Try this prompt:
You are a real estate analyst preparing a comparable sales analysis.
I’m pasting raw sales data from my market. The data is messy — inconsistent formatting, missing fields, different date formats.
[PASTE YOUR DATA HERE]
Please:
1. Clean and standardize the data
2. Calculate for each sale:
- Price per unit
- Price per square foot (if SF provided)
- Cap rate (if NOI provided)
3. Remove any outliers or incomplete records
4. Create a formatted comparison table sorted by sale date (most recent first)
5. Provide a brief summary of market trends based on this data:
- Average and median price per unit
- Cap rate range
- Any notable patterns
Format the final table so I can paste it directly into a report.7. Presentation Decks for Pitches and Approvals
What Anthropic said: “Opus 4.5 powers agents that create PowerPoint presentations with professional polish.”
What this means for you:
Investment committee decks.
Broker opinion of value presentations.
Client pitch materials.
From raw deal information to slide-ready content.
Try this prompt:
You are an investment analyst preparing a presentation for the investment committee.
Create a 10-slide investment committee deck for this acquisition:
PROPERTY
- Name: Sunrise Gardens Apartments
- Location: Tampa, FL
- Units: 48
- Year Built: 1998
- Asking Price: $6.2M ($129,167/unit)
FINANCIALS
- Current NOI: $380,000
- Pro Forma NOI (Year 3): $465,000
- Going-in Cap Rate: 6.1%
- Stabilized Cap Rate: 7.5%
BUSINESS PLAN
- Light value-add: unit interiors, exterior paint, signage
- CapEx budget: $8,500/unit
- Rent increase target: $150/unit
SLIDE STRUCTURE:
1. Executive Summary
2. Investment Thesis
3. Property Overview (location, photos placeholder)
4. Unit Mix & Rent Comps
5. Financial Summary - Current
6. Financial Summary - Pro Forma
7. Value-Add Business Plan
8. Market Overview
9. Risk Factors & Mitigants
10. Recommendation & Next Steps
Keep text minimal. Use bullet points. Make it presentation-ready.Quick aside:
If you’re reading this and thinking...
“I want more of this.”
“I want workflows I can actually use.”
“I want to be around CRE pros who get this stuff.”
There’s one community I recommend.
450+ commercial real estate professionals.
80+ battle-tested AI workflows.
Weekly live calls.
No coding required.
This is the type of stuff we share and talk about in there.
Honestly?
One of the best places to get started if you’re serious about AI in your deals.
Members are saving 10-20 hours per week on underwriting, market analysis, and deal sourcing.
There’s a 7-day free trial.
The Bottom Line
You don’t need to understand AI benchmarks.
You don’t need to care about “SOTA performance.”
You need to know one thing:
Can this model do work that saves me hours?
Based on what Anthropic shipped today...
The answer is yes.
Pick one use case from this list.
Test it today.
Reply and tell me what happened.
Talk soon,
Avi
P.S. - If you know someone still doing underwriting models or investor reports by hand, forward this to them. They’ll thank you.
Join 230+ professionals getting practical AI tactics every week.



This is terrific, thank you. Of these, the closest workflow to one I'm doing already is "Chrome-Based Deal Research (Without Tab Chaos)". I am already (and manually) extracting detailed, unstructured data from numerous (public) sources, then tapping custom GPTs to distill, organize, and output into a spreadsheet.
I'm continually looking for ways to increase the volume and fidelity of the output, allowing me to bring more confidence to my (AI-augmented) analyses.