Does AI work for commercial real estate underwriting?

For parts of it, yes. These jobs graded "works today": T-12 normalization and underwriting assumptions, Reading an OM and first-pass screening, Full acquisition underwrite, Development feasibility and Lease-up reforecast and take-out sizing. On the rest, the best model got most numbers right but missed judgment calls, so a person still has to review it.

Can I trust the numbers an AI model gives me?

More than the judgment. On average the best models got more of the numbers right than the judgment calls. The check models missed most, in rent roll analysis: "Every lease rolls within twelve months; the broker's 'immediate mark-to-market' is the same fact as 100% turnover exposure in year one." (0 of 21 memos got it). Check the reasoning, not just the math.

Why are the judges Claude models?

The first grading used OpenAI models that were also competing, so models graded their own work. We regraded with two Anthropic models that did not compete. PSV builds on Claude, which is why that is in the disclosure.

Were Claude models tested?

Not in version 1. A model cannot fairly judge its own family, so testing Claude needs a non-Anthropic judge panel. That is planned for the next version.

PSV Research · v1, October 2026

What AI actually works in commercial real estate

We tested AI models on 9 everyday CRE jobs and graded each one: works today, works with review, or not ready. Every grade comes from the test results, by a rule we fixed before we saw them.

See the grades How we graded

The short answer

We gave seven AI models nine everyday CRE jobs, three attempts each, and graded every memo against a computed answer key. Result: five jobs work today and four jobs work with review. Across all 189 memos, 31% had nothing wrong in them. The best models usually get the numbers right; more of their misses are judgment calls a seasoned analyst would catch.

Works today5
Works with review4
Not ready0

Disclosure

Pacific Software Ventures (PSV) builds AI systems for commercial real estate firms, so read this with that in mind. Version 1 tests OpenAI models only. The two judges are Anthropic's Claude Fable 5.1 and Claude Opus 5; neither competed, and PSV also builds on Claude. The deals are synthetic, built so every answer can be computed.

The scorecard

Every job we tested, graded

Graded on the best model we tested for each job. The counts are out of three attempts.

Commercial real estate jobs graded by whether AI works today. For each job: the grade, the best model's result and where models most often went wrong. Judges: claude-fable-5-1 and claude-opus-5.
Job	Grade	Best model's result	Where it breaks
Rent roll analysisParse a rent roll, reconcile it and flag what matters	Works with review	GPT-5.6 Luna0 of 3 memos fully right · 100% of numbers · 83% of judgment calls0 of 7 models wrote at least one fully right memo	Every lease rolls within twelve months; the broker's 'immediate mark-to-market' is the same fact as 100% turnover exposure in year one. (0 of 21 memos got this) Suite D has been vacant since January 2026 and is being marketed at a rent well above the building's in-place retail rents. (0 of 21 memos got this)
T-12 normalization and underwriting assumptionsNormalize a T-12 and set defensible going-in assumptions	Works today	GPT-5.6 Terra2 of 3 memos fully right · 100% of numbers · 98% of judgment calls4 of 7 models wrote at least one fully right memo	Three units carry one month free; contractual rent overstates effective rent for those leases. (10 of 21 memos got this)
Reading an OM and first-pass screeningRead a broker OM and decide whether the deal is worth a look	Works today	GPT-6 Astra3 of 3 memos fully right · 100% of numbers · 100% of judgment calls7 of 7 models wrote at least one fully right memo	No check was missed by most memos.
Full acquisition underwriteUnderwrite a deal from the full data room to a decision	Works today	GPT-6 Astra3 of 3 memos fully right · 100% of numbers · 100% of judgment calls6 of 7 models wrote at least one fully right memo	No check was missed by most memos.
Broker opinion of valuePrice an asset from comps and the rent roll	Works with review	GPT-5.6 Terra1 of 3 memos fully right · 100% of numbers · 96% of judgment calls1 of 7 models wrote at least one fully right memo	The T-12 includes $62,000 of insurance proceeds and six months of a departed tenant's rent; both come out of normalized NOI. (1 of 21 memos got this)
Asset management variance and LP letterExplain actuals vs budget, reforecast and draft the LP note	Works with review	GPT-5.40 of 3 memos fully right · 100% of numbers · 89% of judgment calls0 of 7 models wrote at least one fully right memo	The pre-close $6,100 receivable on Unit 211 belongs to the seller under the PSA and is not our bad debt. (1 of 21 memos got this) Suite C paid rent through March as budgeted; its Q1 rent is not run-rate and the April cliff is already in the budget. (3 of 21 memos got this)
Development feasibilityTest yield on cost and whether a site pencils	Works today	GPT-6 Astra3 of 3 memos fully right · 100% of numbers · 100% of judgment calls5 of 7 models wrote at least one fully right memo	The lender sizes the interest reserve to the GC's schedule and requires funded off-site obligations, so the sponsor's equity requirement is understated. (9 of 21 memos got this)
Construction draw reviewCheck a draw request against budget, retainage and lien waivers	Works with review	GPT-5.51 of 3 memos fully right · 100% of numbers · 96% of judgment calls4 of 7 models wrote at least one fully right memo	No check was missed by most memos.
Lease-up reforecast and take-out sizingReforecast a lease-up and size the take-out loan	Works today	GPT-6 Astra3 of 3 memos fully right · 100% of numbers · 100% of judgment calls5 of 7 models wrote at least one fully right memo	No check was missed by most memos.

Works today: The best model we tested got it fully right in at least 2 of 3 attempts and got at least 95% of the numbers right.
Works with review: The best model got at least 90% of the numbers and 80% of the judgment calls right, but most attempts still had something wrong. Usable, if a person checks it.
Not ready: Below that bar. Don't rely on it for this job yet.

Not graded

What version 1 did not test

Lease abstractionNot in this benchmark yet.
IC memo draftingCovered only inside the full-underwrite job, not graded on its own.
Owner and investor outreachNot a judgment task we can key; not graded.
Automated valuation (AVMs), predictive market picks, autonomous investment decisionsNot tested. We don't grade what we haven't run.

Method

How we graded

nine jobs on synthetic deals, each with an answer key computed from the deal's own numbers, so a correct answer exists for every check.
seven models, 3 attempts per job, called through the API with the same prompt and documents, no tools or web access.
A memo is clean only if every number is inside tolerance, every key finding is made, no mistake is flagged by both judges, the decision is right and the memo is complete.
Two judges answer closed yes-or-no questions with a quoted passage as evidence. A mistake counts against a model only when both judges flag it.
The grading rule was fixed on 2026-10-19, before these results were produced, and applied to the best model on each job.
The OM test scores one number: two checks were dropped for every model because the OM alone cannot settle them, so its grade rests mostly on the risk and decision checks.
Limits: three attempts per model is a small sample, the deals are synthetic, and API results differ from the ChatGPT or Claude apps, which add their own tools and instructions.

The two judges agreed on 0.91 by Cohen's kappa (1 is perfect agreement).

Questions

What people ask

Does AI work for commercial real estate underwriting?: For parts of it, yes. These jobs graded "works today": T-12 normalization and underwriting assumptions, Reading an OM and first-pass screening, Full acquisition underwrite, Development feasibility and Lease-up reforecast and take-out sizing. On the rest, the best model got most numbers right but missed judgment calls, so a person still has to review it.
Can I trust the numbers an AI model gives me?: More than the judgment. On average the best models got more of the numbers right than the judgment calls. The check models missed most, in rent roll analysis: "Every lease rolls within twelve months; the broker's 'immediate mark-to-market' is the same fact as 100% turnover exposure in year one." (0 of 21 memos got it). Check the reasoning, not just the math.
Why are the judges Claude models?: The first grading used OpenAI models that were also competing, so models graded their own work. We regraded with two Anthropic models that did not compete. PSV builds on Claude, which is why that is in the disclosure.
Were Claude models tested?: Not in version 1. A model cannot fairly judge its own family, so testing Claude needs a non-Anthropic judge panel. That is planned for the next version.

Next step

Find out which of these jobs AI can take off your desk.

Ten minutes. An honest read on where you stand and which work is worth handing to AI, if any.

Get your free AI assessment