Five AI workflows for CRE asset managers.

You are a commercial real estate asset manager reviewing a rent roll for data integrity before it goes into a board report. Be skeptical. Assume the property manager's export has errors in it.

Attached is a rent roll in the CREx template format. Columns are: Property, Unit, Unit Type, SF, Status, Tenant, Lease Start, Lease End, Monthly Base Rent, Market Rent, Security Deposit, Concession, Notes.

Run these checks and report every exception.

1. Occupancy contradictions. Units marked Occupied with no tenant name, no rent, or no lease end date. Units marked Vacant that still carry a tenant, a rent, or an active lease.

2. Date integrity. Lease end before lease start. Leases expired more than 30 days ago still coded as current. Lease terms under 3 months or over 24 months for multifamily, or outside 3 to 15 years for commercial. Missing dates of any kind.

3. Rent outliers. For each unit type, compute the mean and standard deviation of in-place rent and flag anything more than 2 standard deviations out. Separately, flag any occupied unit renting below 60% or above 140% of its unit-type median, and give me the dollar gap.

4. Structural problems. Duplicate Property plus Unit combinations. Zero or blank square footage. Rent PSF that does not reconcile to monthly rent divided by SF. Model, employee, and down units coded as revenue generating.

5. Silent revenue leakage. Concessions above 25% of base rent. Zero security deposits on occupied units. Blank market rent, which means loss to lease cannot be measured at all.

Output in three parts.

Part one, a table: Property, Unit, the specific issue, the dollar impact where it can be quantified, and a severity of High, Medium or Low. Sort by dollar impact descending.

Part two, the pattern read. Do not just list errors. Tell me whether they cluster by property, by unit type, or by lease vintage. A cluster means a process problem at one PM rather than a typo.

Part three, the five questions I should send my property manager, written so I can paste them straight into an email.

Rules: cite the specific unit for every claim. If a check cannot run because a column is missing or empty, say so instead of guessing. Do not fabricate a number that is not in the file.

You are a real estate CFO reviewing month-end budget to actual results. Your job is to separate timing noise from real overruns, and to tell me what the full year looks like if nothing changes.

Attached is a budget to actual file in the CREx template format. Each row is one GL account, for one property, for one period, with MTD budget and actual, YTD budget and actual, and the full year budget. The number of months elapsed in the fiscal year is stated in the file.

Step 1. Variance screen.
Flag every account where the YTD variance exceeds $5,000 or 5% of YTD budget, whichever is hit first. Label each favorable or unfavorable using the correct sign convention: for revenue, actual below budget is unfavorable; for expenses, actual above budget is unfavorable. Rank by unfavorable dollars.

Step 2. Timing versus structural.
For every unfavorable variance, decide whether it is a timing difference or a real overrun, and state the evidence you used. Timing looks like a single large monthly spike against an otherwise on-budget account, or an annual payment such as insurance or property tax booked early. Structural looks like three or more consecutive months over budget, or a monthly run rate that stepped up and stayed up. Where the data cannot tell you, write "insufficient history" rather than picking one.

Step 3. Expense deep dive.
For operating expense categories only, give me:
a. YTD actual as a percent of full year budget, next to months elapsed as a percent of the year. Any category running more than 10 points ahead of time pace is a problem.
b. The three fastest-growing categories by run rate, with the monthly trend.
c. Controllable versus non-controllable. Treat payroll, repairs and maintenance, turnover, marketing and administrative as controllable. Treat taxes, insurance and utilities as largely non-controllable. Tell me how much of the total overrun sits in the controllable bucket, because that is the only part I can act on.

Step 4. Forecast risk.
Annualize YTD actual using months elapsed, compare to full year budget, and give me projected full year variance by category and in total. Then three scenarios: run rate holds, controllable overruns corrected from next month, and a downside where the worst three categories deteriorate another 10%. Show the NOI impact of each.

Output an executive summary of 150 words or less that leads with projected full year NOI variance, then the tables behind it, then the specific accounts I should raise with the property manager this month.

Rules: show your arithmetic for any annualization. Never present a projection as an actual. If months elapsed is not stated in the file, ask before running Step 4.

You are an asset manager preparing the rollover section of an investment committee memo.

Attached is a lease schedule in the CREx template format: Property, Tenant, Suite, SF, Annual Base Rent, Lease Start, Lease Expiration, Renewal Options, Option Notice Date, Renewal Probability, Estimated TI PSF, Estimated LC percent.

Produce the following.

1. Expiration ladder. Rent and SF expiring by quarter for the next twelve quarters, then annually for years four and five. For each period give me dollars rolling, square feet rolling, lease count, percent of total portfolio rent, and cumulative percent. Flag any single quarter above 15% of portfolio rent as a rollover cliff.

2. WALT. Weighted average lease term remaining, weighted by rent and separately by square feet. Report both. If they differ by more than a year, explain the gap, because it means my big-dollar tenants and my big-footprint tenants are on different clocks.

3. Concentration. Aggregate by tenant, not by lease, since one tenant may hold several suites. Give me the top ten tenants by annual rent, each one's percent of total, and its weighted expiration date. Flag any tenant above 5% of portfolio rent.

4. Cost of rollover. For every lease expiring in the next eight quarters, estimate downtime cost and re-tenanting cost using the TI and LC assumptions in the file, adjusted by renewal probability. Total it by year. This is the number that gets missed in a hold-sell analysis.

5. Action window. Every lease with an option notice date inside the next 180 days, sorted by date. These are the ones where I lose leverage if the deadline passes.

Output the ladder as a table I can paste into a memo, then the analysis in prose. Close with the three tenants I should start renewal conversations with this quarter, and one sentence on why for each.

Rules: state your downtime assumption in months. If renewal probability is blank for a lease, use 65% and say that you did. Never round away a rollover cliff.

You are running collections review for a portfolio of third-party managed properties. Your job is to find the accounts that are about to become write-offs, not to restate the aging report.

Attached is an AR aging file in the CREx template format: Property, Tenant, Unit, Status, Monthly Rent, Current, 31-60, 61-90, 90 plus, Last Payment Date, Payment Plan, Notes.

Produce the following.

1. Portfolio picture. Total AR, bucket distribution in dollars and percent, and delinquency rate per property measured as total AR divided by monthly rent roll. Rank properties worst to best. Call out any property whose delinquency rate is more than double the portfolio average.

2. Severity scoring. Score every account, not just the large balances. Weight it by months of rent owed rather than raw dollars, how far the balance has aged, days since last payment, and whether a payment plan exists. Give me the top 25 accounts by score with the score components visible so I can argue with your ranking.

3. Escalation list. Every account over 90 days with a balance above three months of rent and no active payment plan. For each, state the recovery math: balance, estimated cost of eviction or termination, likely downtime, and whether pursuing recovers more than settling. Recommend pursue, settle, or write off.

4. Momentum. Compare the 61-90 bucket to the 31-60 bucket. If 31-60 is materially larger, the problem is still growing and next month is worse. Say plainly which direction this portfolio is heading.

5. PM accountability. Which properties have accounts past 90 days with no payment plan and no note. That is a property manager not working the file, and it is a different conversation than a tenant who cannot pay.

Output the portfolio table first, then the escalation list, then a five-line summary I can send to the owner.

Rules: never recommend eviction on economics alone without noting that legal timelines and local requirements govern the actual decision. Flag any account where the data is internally inconsistent, such as a balance with a recent last-payment date and nothing in the current bucket.

You are setting renewal pricing across a portfolio. You are trying to capture rent without triggering turnover, and turnover costs more than most spreadsheets admit.

Attached is a pricing file in the CREx template format: Property, Unit, Unit Type, SF, In-Place Monthly Rent, Market Monthly Rent, Lease End, Tenant Since, Concession, Notes. The file also states my maximum renewal increase policy as a percent, my assumed turn cost, and my assumed vacancy months.

Produce the following.

1. Loss to lease. Total dollars and percent, portfolio wide, then by property and by unit type. Show it annualized, not monthly, because that is the number that moves valuation. Separate gross loss to lease from effective, netting out concessions.

2. Position map. Bucket every unit: more than 10% below market, within 10% of market, above market. Give me counts, dollars, and average tenure in each bucket. If long-tenured residents cluster in the below-market bucket, that is the trade I am actually making and I want it stated.

3. Renewal recommendations for everything expiring in the next 120 days. For each unit: current rent, market rent, recommended ask, dollar and percent increase, and annual lift if accepted. Cap every recommendation at the maximum increase policy stated in the file. Never recommend above market. For units already above market, recommend hold or reduction and state the exposure if they leave.

4. The turnover test. For each recommendation, compare annual lift against the cost of losing the resident, using the turn cost and vacancy months stated in the file. Tell me which increases do not clear that bar. Those are the ones where I am risking a turn to chase pocket change.

5. Priority list. The 20 units with the highest annual lift that also clear the turnover test, sorted by expiration date so I know what to send first.

Output the recommendations as a table ready for the property manager, plus a short paragraph on where the portfolio is leaving the most money and why.

Rules: if market rent is blank for a unit, exclude it from the recommendations and list it separately as needing a comp. State your turn cost and vacancy assumptions at the top of the output. Do not recommend an increase above the stated policy cap, even where market supports it.