Updated 3 November 2026.
Short answer: on five real offering memorandums, Titleman Underwriting (fast screening) found all 23 key issues in our answer key and wrote 11 to 15 broker questions per memorandum, ordered by value and tied to a page and a figure. A general AI model setup found 9.5 of the 23 and wrote longer, generic question lists. It was faster on some memorandums and no worse on arithmetic checks. This is a small experiment, not a benchmark. For the checks themselves, see how to check the numbers in an offering memorandum.
The table
| Incoming offering memorandum, 5 real listings | Analyst by hand | Person + general AI model | Titleman Underwriting (fast screening) |
|---|---|---|---|
| Key issues found | not run | 9.5 of 23 | 23 of 23 |
| Wrong numbers | not run | 0 | 0 |
| Questions for the broker | not run | 17 to 21, generic, unordered | 11 to 15, ordered, with page and figure |
| Your time | about 120 min* | about 30 min* | about 20 min* |
| Machine time | none | median 47 min | median 83 min |
*Estimates, not measured. Footnote: experiment on five offering memorandums from open broker listings, answer key fixed before the runs, one run per cell. The "person + general AI model" column is the same engine with Titleman features switched off and a generic prompt, not a person at a chat screen. The prompt is published below. Times for people are estimates; machine times include queueing. Small sample: read it as a measured experiment, not a benchmark.
Per memorandum
Key issues found, Titleman against the general setup:
| Memorandum | Titleman | General setup |
|---|---|---|
| 34-unit multifamily, Kansas | 5 of 5 | 3.5 of 5 |
| Multifamily, Texas | 5 of 5 | 0 of 5, stopped at a parcel check |
| Multifamily, Massachusetts | 4 of 4 | 1 of 4 |
| 37-unit multifamily, Georgia | 4 of 4 | 3 of 4 |
| 72-unit multifamily, Georgia | 5 of 5 | 2 of 5 |
Where the general model is not worse
- It was faster where nothing queued: 47 minutes against 101 on the Kansas memorandum.
- It checked rounding exactly, and caught a page in the Georgia memorandum that describes another city.
- It flagged a zoning question on the 37-unit memorandum.
- It produced long due-diligence lists (litigation, environmental, contract terms) in one pass.
- For a 30-minute first pass it is a fair tool.
Where Titleman is ahead, and where it is weak
- Ahead on tie-out depth: it reconciles the rent roll, the T-12 and the headline numbers line by line. That caught vacant units carried at market rent, a property tax formula that the figures do not support, a loan with no terms, and cap rate and rent multiple bases that disagree. The general setup found the loud errors and said "ties" on some that do not.
- Ahead on the questions: ordered by value, each with its page and figure, plus a verdict and its own value range.
- Weak on time: 70 to 100 minutes of machine time per memorandum, so it is not instant.
- Weak on coverage: 12 of 17 secondary items found. It missed comps per unit, a bonus loft and demographics.
Method
- Five real offering memorandums from open broker listings. No client documents were used.
- The answer key was written before any run: rent roll, offering memorandum and T-12 mismatches, occupancy, loan term, tax, risks, and the questions an experienced underwriter would ask. The file hashes were fixed with it.
- The same prompt for both AI columns:
Attached is an offering memorandum for a N-unit apartment property at ADDRESS. Please review it for me. Summarize the deal, check the numbers for errors or inconsistencies, tell me what looks risky or missing, and write a list of questions I should send to the broker.
- Scored by item against the key: issues found, wrong numbers (recomputed from the figures each answer quoted), and questions that match the key.
- The analyst column is the catalog estimate of two hours for a light screening, not a measured run.
- One run per cell, five memorandums. One general-setup answer showed comparables from another state in an empty value panel; it is counted as a miss, not as a wrong number.
FAQ
Can an AI model screen an offering memorandum?
Yes, as a first pass. In this experiment a general AI setup found 9.5 of 23 key issues in about 47 minutes of machine time. Titleman found 23 of 23 in about 83 minutes.
What did the general model miss?
Mostly tie-out issues: vacant units carried at market rent, a property tax line the figures do not support, and a loan with no stated terms. More on where a general chat model stops: can ChatGPT underwrite a commercial real estate deal.
How do I send a memorandum for screening?
Forward the offering memorandum. It comes back screened, with a list of questions for the broker, like this broker questions sample. Request a demo to try it on one from your inbox.