Investing & underwriting

Pressure-test a real estate investment thesis with a Claude research skill that runs five expert personas in parallel and grades every source

The AI for CRE Collective packaged STORM, a research method from Stanford, into a reusable Claude skill and ran the skill on one question: are data centers a safe commercial real estate bet, or is stranded-asset risk hiding behind the leasing numbers? From one plain prompt, the skill spins up five sub-agents that research the web in parallel, each from a different lens: a practitioner who knows what breaks in practice, an academic who trusts only real data, a skeptic who builds the case against the popular narrative, an economist who follows how the deals are financed, and a historian who compares the cycle with earlier ones. The skill then maps where the five lenses contradict each other, writes one synthesis, and peer-reviews its own work before producing a single HTML report: a 60-second read, key findings, a hidden-connection section, and a source ledger that labels each source as corrected, confirmed, or unverified. The lenses ship in generic business language, so the article recommends rewriting the lenses in the skill file for real estate: an MAI appraiser who wants closed-transaction comps and cap rates, a skeptic building the bear case from foreclosures and distress sales, a lender seat tracking debt yields, and a historian focused on real estate cycles. Expect many web-access approvals during a run.

At a glance

Tools used
ClaudeClaude Skills
Category
Investing & underwriting
Audience
investor, broker
Difficulty
No-code setup
Source
How We Turned Stanford's STORM Method Into a Repeatable CRE Research Skill
PLOT Daily Editorial Desk (AI for CRE Collective) · Article · Jul 24, 2026
First step
Write the one investment question you want pressure-tested this month, such as whether to add a property type or market to your buy box, in one sentence.

Compliance note: A STORM report is research to pressure-test a thesis, not underwriting or investment advice. Read the source ledger and verify any claim marked unverified before the claim reaches an investment memo, an investor, or a client.

Suggested prompt

Written by this directory, inspired by the source. Replace the [bracketed] parts.

Run a STORM research on [property type or market]: build an investment thesis with the upsides, the downsides, the hidden risks, and what people are not talking about. Use five researchers working in parallel as separate sub-agents: an operator who knows what breaks in practice; an MAI appraiser who trusts only closed-transaction comps, cap rates, and published market reports; a skeptic who builds the bear case from foreclosures, cap-rate expansion, and distress sales; a lender who tracks cap rates, debt yields, and how deals are financed; and a historian who compares this cycle with past real estate cycles. Then map where the five disagree, write one synthesis, review your own work for weak claims, and deliver one HTML report with a 60-second summary, key findings, and a source ledger marking each source as confirmed, corrected, or unverified.

Share on X Read the source

← All use cases