An AI analyst can take the first read of every offering memorandum (OM) off your desk: it checks the numbers, flags the errors and drafts the broker questions. It does not take the bid decision, and it is slower on the clock than a person reading a single OM. What it changes is how many OMs get a full first read in a week.
Published by Реалист, which makes the tool described. Updated 2 November 2026.
A week of OM screening, today and with Реалист
Figures marked "experiment" come from our test. Figures marked "illustrative" are arithmetic on assumptions you should replace with your own. We have no named client case for this yet, so the second column is what Реалист can do, not what a customer reports.
| Мера | Junior analyst today (illustrative) | With Реалист |
|---|---|---|
| OMs screened per week | Whatever fits in the hours: at about 2 hours each, 20 OMs is a 40-hour week | Illustrative: 30 inbound OMs can all start the same day, because runs go side by side |
| Hours per OM, person | About 2 hours by hand (our estimate in the experiment, not a timed study) | About 20 minutes of human review (estimate in the experiment, n=3) |
| Machine time per OM | Нет | 69 to 101 minutes (experiment, n=3) |
| Broker questions per OM | Depends on the analyst and the day | 13 to 14, tied to pages and figures (experiment, n=3) |
| Who decides | The analyst proposes, the associate reviews | You decide; the analyst's hours go to the questions that change the bid |
If 30 OMs arrive in a week, the illustrative human time is about 10 hours of review instead of about 60 hours of reading. That is extrapolated from three OMs. Treat it as a hypothesis to test on your own pipeline.
What the experiment measured
We ran three real OMs through Реалист and scored each against a list of key errors we had identified by hand first. This is an experiment, n=3, not a benchmark.
- Key errors found: 5 of 5 on the first OM, 4 of 4 on the second, 5 of 5 on the third.
- Broker questions: 13 to 14 per OM, covering every key issue we had listed.
- Time: 69 to 101 minutes of machine time per OM.
The weak spot: the engine is slow per OM. A person can skim an OM in less time than that, and a single urgent OM is not where this helps. We also compared it with a general-purpose AI prompt on the same engine, but one of the three comparisons stopped at an address check and is not counted, so we do not publish a head-to-head number. We have not measured a person using a consumer chat tool.
Where it fits in a screening week
- Forward the OM. The engine reads it against records and your buy box.
- Open the reply: errors, gaps, and a drafted list of broker questions.
- Your analyst or associate spends the saved hours on the calls and the deals that pass.
The wider tiered process is in how funds screen 30 deals a week, and the OM steps are in how to analyze an offering memorandum with AI.
What an AI analyst does not replace
- Judgment on the bid. Price, structure and risk appetite stay with you.
- The broker relationship. The questions are drafted; someone still has to ask them and read the tone of the answer.
- A licensed appraisal, a title search or an engineer's report. A screen is not any of these.
- Training. A junior analyst becomes an associate. A tool does not.
- Checking the output. Reports name what they could not find, and a person should read those gaps first.
Вопросы и ответы
Can an AI analyst replace hiring a junior analyst at a real estate private equity firm?
It can take over the first read of each OM. It does not replace the person who makes the decision. If the open role is mostly screening, test the tool on a few OMs before you decide.
How many OMs can an AI screen in a week?
Not measured for a full week. Runs go side by side, so the limit is your review time.
How do I test it?
Send the OMs you already screened and compare the error list and questions with your analyst's notes.
Запросить демо and bring last month's OMs.