The AI Analyst: How AI Can Help Screen Deals
Meet my Quant, updated in April 2026
Do you remember the scene from The Big Short where Jared Vennett introduces his quantitative analyst, the award-winning math specialist who calculated that an 8% default rate would result in MBS Armageddon?
Today, I’ll introduce you to your very own quant that will help you with due diligence on real estate deals. It’s available right now, costs less than a dinner out, and will never expense a bottle of Scotch to the deal.
The downside, you ask? It’s a tool, it makes mistakes, and the context is important (this tool provides none). The biggest value (to me) is the questions it generates.
In this post, I’ll show you what AI can do to help you extract information from pitch decks and subscription docs.
If you invest in real estate deals, you’ll find our series on how to read financial statements very helpful. This is my favorite part
You’ll find Part 1, Part 2, Part 3, and Part 4 worth a read as well.
DISCLAIMER: This is not a replacement for due diligence. AI is a tool that accelerates analysis, surfaces risks, and improves efficiency. Every output should be independently reviewed and validated before you make any investment decision. Consult qualified professionals (AI is not one). Nothing here is legal, tax, or investment advice.
Pick your tool
The number one question I get from readers is: which AI should I use?
The short answer: use whichever one you’re comfortable with.
ChatGPT is the easiest entry point. Clean interface, good all-around capabilities, brief responses. The cons: it hallucinates more than I’d like, tends to agree with whatever you tell it, and the free plan has a low token limit.
Claude is my personal favorite at the moment. It handles long documents well, thinks more carefully before answering, and produces more nuanced output. It talks a lot, which can be annoying. If you use Excel, the plug-in is incredible.
NotebookLM is Google’s deliberately constrained model. I call it the dumb one (and I mean that as a compliment). It only works with documents you upload, which means it cannot hallucinate information from outside your sources. For reviewing PPMs and Operating Agreements, this constraint is a great feature. It is the least likely to make things up.
Before you type a single prompt
Truth be told, prompting isn’t as important as it was before. Why? Because LLMs are getting smarter by the day, and can figure out what you are trying to get to.
Still, most people type something like: “Tell me if this is a good investment.” AI cannot answer that question. What it will do is summarize the sponsor’s narrative back to you, dressed up as analysis. That is not analysis. That is a very fast way to get sold a deal.
A good prompt has four components: