Real estate financial modelling often involves bringing together a large number of assumptions — from development costs and rental income to financing, exit values and returns. In a recent Cambridge Finance Mastermind session, Maria Wiedner worked through a development case study to demonstrate how these inputs can be structured into a development appraisal and how AI tools can support the modelling process.
The session also highlighted an increasingly important skill for real estate professionals: knowing how to use AI effectively while still being able to critically assess the model it produces.
Starting with the Right Model
The first step in any financial modelling exercise is understanding what type of model the project requires.
For the case study, the focus was a property development involving construction costs, rental assumptions, financing and an eventual exit. The appropriate approach was therefore a development appraisal, with the model structured around three key areas:
- Development and construction
- Operations and income
- Exit and returns
The objective was to understand the relationship between the costs required to deliver the development and the returns generated when the completed asset is sold.
Establishing the Key Inputs
Before building the model, the key assumptions need to be identified and organised.
The case study included assumptions around the number and type of beds, rental levels, rental growth, operating expenditure, land costs, construction costs, professional fees, contingency, development costs and financing.
For example, the model considered different accommodation types and rental assumptions, alongside operating costs and an expected exit date.
Structuring these inputs clearly is important because the assumptions ultimately drive the development appraisal and the resulting investment returns.
What Happens When AI Builds the Model?
One of the most interesting parts of the session was testing how an AI tool could help build the model.
Microsoft Copilot was used to interpret the case study and generate elements of the development appraisal. Within a short period, it produced calculations covering areas such as rental income, operating expenditure, exit valuation and development costs.
This demonstrated how AI can significantly reduce the time required to create an initial model structure.
But there was an important caveat.
An AI-generated model still needs to be checked.
As Maria highlighted during the session, AI-generated outputs can look convincing even when an assumption, formula or calculation needs further review. The modeller therefore needs to understand the underlying mechanics rather than simply accepting the output.
Understanding the Numbers Behind the Model
The session involved checking the AI-generated calculations against the original case study assumptions.
This included reviewing:
- Number of beds and accommodation types
- Rental assumptions and rental growth
- Operating expenditure
- Stabilised NOI
- Capitalisation rate
- Exit valuation
- Disposal costs
- Senior debt repayment
- Equity proceeds
For example, the exit valuation was linked to the stabilised NOI and capitalisation rate. The discussion emphasised the importance of understanding how each figure is calculated and where the underlying assumption comes from.
This is particularly important when using AI because a model can appear technically complete while still containing assumptions that require professional judgement.
Modelling the Development Cost Profile
Another important part of the discussion was the timing of development costs.
Rather than simply showing the total development cost, a financial model needs to consider when those costs occur.
The case study used an S-curve to model construction expenditure over the development period. This helped demonstrate how construction costs can build up over time rather than being incurred evenly throughout the project.
Other costs may follow different patterns.
For example, professional fees and development management fees may need to be treated differently depending on when the underlying services and payments occur. The session explored separating these costs rather than automatically applying the same timing assumption to all development expenses.
This is a good example of where real-world modelling judgement becomes important.
Sensitivity Analysis
The session also discussed sensitivity analysis as an important part of a development appraisal.
A development model is based on assumptions, and changes to those assumptions can materially affect the resulting returns.
Sensitivity analysis allows the modeller to examine how the outcome changes when key assumptions change.
Rather than looking only at the base case, investors and development professionals can use sensitivity analysis to understand the potential impact of changes to variables such as:
- Rental assumptions
- Exit value
- Development costs
- Financing costs
- Timing
The purpose is not simply to produce another table of numbers, but to understand which assumptions are most important to the investment outcome.
AI as a Modelling Assistant, Not a Replacement for the Modeller
Perhaps the biggest takeaway from the session was the role AI can play in financial modelling.
AI can help with:
- Interpreting a case study
- Structuring a model
- Generating formulas
- Producing an initial development appraisal
- Building supporting calculations
- Identifying potential areas for review
However, the modeller still needs to understand the fundamentals.
As the session demonstrated, AI can produce a sophisticated-looking model very quickly. The challenge is determining whether the model is actually correct.
That requires an understanding of development appraisal mechanics, cash flows, assumptions, valuation and the timing of costs.
The Future of Real Estate Financial Modelling
AI is changing how financial models can be created and reviewed. For real estate professionals, the opportunity is not necessarily to replace traditional modelling skills, but to combine those skills with new AI capabilities.
A strong workflow could involve building a model manually, using AI to accelerate or challenge the process, and then comparing the outputs and checking the underlying calculations.
The Mastermind session demonstrated this approach in practice: use AI to move faster, but use financial modelling knowledge to decide whether the result makes sense.
That combination of technical modelling skills, commercial understanding and AI literacy is becoming increasingly relevant for professionals working across real estate investment, development and finance.
