Financial modelling is no longer just about building spreadsheets. Today, AI tools like Claude can help real estate professionals structure investment cases, create dashboards, and even generate development appraisals faster than ever before.
In this webinar, Maria Beadner walked through a live real estate case study to demonstrate how AI can support investment analysis and stress testing — while also explaining why human judgement remains essential.
Why Most Financial Models Fail
One of the key points discussed during the session was that many analysts build models from the wrong direction.
Traditionally, analysts open Excel, input assumptions, calculate outputs, and then hope the model tells a compelling story. But a model only performs arithmetic — it does not communicate an investment thesis.
Before opening Excel, investors should first define:
- What decision needs to be made
- What question the investment committee needs answered
- Which KPIs matter most
- Which variables need to be stress tested
This “output-first” approach creates clearer investment cases and more effective dashboards.
The Case Study: Gate House, London
The webinar focused on a real investment opportunity known as Gate House, a freehold office building located in Wentworth, London.
Key details included:
- Asking price: £6 million
- Refurbished office completed in 2024
- Existing occupier: Spacemate
- Annual income: approximately £52,000
- Close to retail and public transport
- Potential for residential conversion
Although the property was technically occupied, the income generated was extremely low relative to the purchase price. This immediately raised concerns about whether the asset worked as a traditional income-producing investment.
The discussion quickly shifted toward a more likely strategy: residential conversion.
Using Claude to Build the Investment Framework
Rather than starting in Excel, Maria demonstrated how Claude can help structure the entire investment case first.
The AI prompt used focused on designing:
- The investment committee dashboard
- KPI hierarchy
- Sensitivity analysis structure
- Cash flow layout
- Stress testing outputs
Example prompt:
“I need to design the output dashboard for a residential conversion development appraisal. The audience is an investment committee. What KPIs should appear, in what order, and what should go into the sensitivity table?”
Claude was then provided with the property brochure and context surrounding the opportunity.
The result was a surprisingly strong first draft of a development appraisal dashboard and investment structure.
What AI Did Well
The AI-generated output included:
- Investment dashboard layouts
- Development appraisal structures
- Sensitivity tables
- Profitability metrics
- IRR calculations
- Profit on cost analysis
- Development assumptions
- Scenario testing
From a presentation perspective, the output was highly effective.
The dashboard clearly highlighted:
- Profitability
- Equity returns
- Build costs
- Exit values
- Sensitivity scenarios
This demonstrated how AI can dramatically speed up the initial modelling process.
Where AI Still Falls Short
While the structure looked impressive, an important warning was repeated throughout the webinar:
AI-generated financial models are not automatically reliable.
Claude was able to generate assumptions and calculations quickly, but several limitations became clear:
1. Static Assumptions
The model lacked flexibility.
For example:
- Build periods did not dynamically adjust
- Sales timelines were fixed
- Cash flow timing was rigid
- Sensitivities did not fully flow through the model
2. Unrealistic Outputs
Some assumptions and numbers required manual verification.
AI can:
- Invent market assumptions
- Miscalculate timelines
- Apply unrealistic build costs
- Create inaccurate financing structures
3. Lack of Commercial Judgement
AI cannot yet connect all the commercial dots.
It does not truly understand:
- Market sentiment
- Local demand
- Leasing risk
- Development complexity
- Planning uncertainty
This is where experienced analysts remain essential.
Stress Testing the Deal
The most important part of the exercise was identifying which variables actually mattered.
For this residential conversion case, the key stress testing variables were:
Gross Development Value (GDV)
The future sale price per square foot had the biggest impact on returns.
Build Costs
Reducing construction costs significantly improved profit on cost.
Purchase Price
The £6 million asking price appeared difficult to justify under realistic assumptions.
The analysis suggested that:
- Lower acquisition pricing would improve viability
- Higher sales values were necessary for attractive returns
- Tight margins created significant development risk
The Importance of Dashboard Design
A major takeaway from the webinar was that:
“The dashboard is the argument.”
Investment committees do not want to read through complex spreadsheets.
They want:
- Clear KPIs
- Structured outputs
- Simple sensitivities
- Fast visibility into risk and return
AI tools like Claude can help professionals create cleaner and more structured investment presentations — especially during the early stages of analysis.
Key Takeaways from the Webinar
1. Start With the Output
Before building a model, define:
- The investment question
- The decision required
- The key metrics
- The stress testing variables
2. Use AI for Structure
Claude can help generate:
- Dashboards
- Model skeletons
- Sensitivity tables
- Investment summaries
3. Human Expertise Still Matters
AI is not a replacement for financial modelling expertise.
Professionals must still:
- Verify assumptions
- Fix model logic
- Test flexibility
- Validate outputs
- Apply market judgement
4. Credibility Comes From Consistency
Strong modelling is not just about calculations.
It is about:
- Understanding assumptions
- Building logical structures
- Creating clear investment arguments
- Presenting reliable outputs
Final Thoughts
AI is rapidly changing the way real estate professionals approach financial modelling and investment analysis.
Tools like Claude can dramatically improve efficiency and accelerate early-stage underwriting. However, the quality of the final investment decision still depends on the analyst’s understanding of real estate fundamentals, market dynamics, and model logic.
The future is likely to combine both:
- AI for speed and structure
- Human expertise for judgement and credibility
As Maria highlighted throughout the session, the goal is not simply to build a spreadsheet — it is to build a clear and defensible investment case.
