How I Built a Real Estate DCF Web App with AI in Five Minutes

Over the past year, we’ve been experimenting heavily with (and successfully using) AI coding tools to push the boundaries of what’s possible in commercial real estate. We’ve used v0 to quickly scope frontend designs, and we’ve spun up various web apps using tools like Replit, Bolt.new, Lovable, and Cursor to support everything from content creation to Accelerator member Q&A to financial modeling support.

Since LLMs came out, there is one particularly challenge that has consistently been rolling around in my head: How do we make it easier for a digital coworker – a concept I’m tackling head-on in my day job at CRE Agents – to perform real estate financial analysis?

As you well know, real estate professionals rely heavily on Excel for modeling. But AI tools today aren’t well-equipped to read, navigate, and manipulate Excel files. Despite how powerful Excel is for humans, it’s a friction point for digital coworkers.

So, I started exploring alternatives and one thought came up:

“I wonder if I could turn an Excel model into a real estate financial model web app using Lovable?”

I decided to find out.

The Experiment: From Excel to Web App Using AI

I began with a basic real estate DCF model I’d built in Excel. I chose it because it has straightforward inputs (blue font), relatively simply calculations (black font), and a few key outputs like unlevered and levered IRRs.

Rather than jumping straight into prompting Lovable, I took an extra step to set up ChatGPT as my prompt assistant. I didn’t just dump screenshots and hope for the best – I made sure ChatGPT understood the context of the model.

Here’s the exact setup I used:

“I want to build a web app that will replicate this basic real estate DCF model. I want to use a no-code solution (e.g., Lovable or Replit+V0) to pull this off. You will be my prompt assistant (or code assistant if we go with Replit), but to start allow me to share first the model and then the logic behind the model. Describe the details of this model.

Note that blue font cells are required inputs — these will be entered by the user (or digital coworker eventually!). Black font cells are either labels or calculations — do your best to differentiate based on context what are calculations and what are labels.

Then I’ll share the logic and we’ll see if you’re correct.”

I also included two screenshots of model:

• The layout of the Excel model (showing inputs, calculations, and outputs).
• The formulas view (using Ctrl + ~), so it could see how calculations flowed through the model.

ChatGPT then analyzed the structure, identified user inputs, calculations, and labels, and summarized its understanding back to me. Then, I asked ChatGPT to draft an initial prompt for Lovable that would recreate the model as a web app.

• Looking for formal training on AI in CRE (and real estate financial modeling more broadly)? Check out our A.CRE Accelerator Program and/or our AI for CRE Course.

One Prompt, Five Minutes, and a Working Prototype

Honestly, I expected this would take multiple prompt iterations. But on the very first try, after waiting about five minutes for the AI to write the code, Lovable produced a working prototype.

• Here’s the working web app: See the live web app here (you’ll need a free Lovable account to view)

The app allows users to:

• Enter basic deal assumptions (purchase price, financing terms, operations, etc.)
• Calculate returns instantly
• View full cash flows
• Export results as a CSV

It’s a fully functional real estate DCF tool, built almost instantly, without having to write a single line of code myself.

Important Caveats: Production Apps Require Much More

Of course, if this were meant to be a production-grade web application, much more work would be required, including:

• Rigorous quality control to ensure calculation accuracy.
• Iterative improvements to enhance UX, design, and functionality.
• Security and data privacy measures.
• Scalability for handling multiple users and large datasets.

This wasn’t about replacing that careful, professional process.

Instead, it was a demonstration of what a non-technical professional can now accomplish with today’s AI tools. And how much faster ideas can move from “concept” to “prototype.”

Watch: Building the Real Estate DCF Web App in Five Minutes

If you’d like to see exactly how this process unfolded, from using ChatGPT as a prompt assistant to launching a working web app on Lovable, I recorded a quick walkthrough video showing each step:

In the video, you’ll see:

• How I structured my initial ChatGPT prompt.
• How Lovable interpreted the prompt and built the first version of the app.
• A live demo of the working web app itself, calculating returns and exporting cash flows.

It’s a simple but powerful demonstration of what’s possible today for non-technical real estate professionals using the AI tools available to us.

Final Thoughts

This exercise reinforced something I deeply believe: The line between “technical” and “non-technical” is disappearing faster than anyone realizes. And furthermore, those who understand the problems, irrespective of their technical capability, will be best equipped to solve them in an AI-enabled future.

If you know what you want, and you know how to collaborate with AI, you can prototype it (and eventually build it) yourself.

It’s an exciting time to be a commercial real estate professional!