Skip to content
All resources

Blog

How to summarize a commercial real estate offering memorandum and data room with AI (and check the OM against the room)

by Daniel Leong, GTMUpdated

What's the best AI tool for summarizing CRE offering memorandums and data rooms? Compare the options, check the OM against the room, and copy our template.

Guide cover reading “Data room”: summarize an offering memorandum and check it against the data room

Short answers

What's the best AI tool for summarizing commercial real estate offering memorandums and data rooms?

It depends on the job. If the seller's virtual data room has built-in AI, it helps you search and ask questions inside that room; a general assistant such as ChatGPT or Claude can summarise documents you are allowed to share, but cross-checking them depends on how each analyst prompts it; and a CRE platform fits when the summary has to tie back to every source document, stay under your firm's control and reach investment committee. Many deal teams use more than one, so pick by job rather than by brand.

Can you upload data room documents to ChatGPT or Claude?

Only if your NDA and your firm's policy allow it. Many NDAs limit who may see the materials and for what purpose, so check the permitted use clause first, use a firm-managed business account rather than a personal one, and keep the deal's working files in your firm's own document library.

How do you check an AI summary of an offering memorandum?

Treat the OM as the seller's story and verify each headline figure (in-place NOI, occupancy, WALT, rent growth) against the rent roll, T-12 and leases in the data room, and later against tenant estoppels. A good AI summary makes that easy by citing the source behind every figure and saying plainly when a document is missing.

How to summarise a commercial real estate offering memorandum with AI

An offering memorandum is a marketing document. It is written by the seller's broker to present the asset in its best light, which is exactly why a buyer's summary should not simply shorten it. A useful buyer-side OM summary pulls out eight things, each with the page it came from:

1. Deal snapshot
Asset, location, size, asset type, asking price or guidance, price per square foot or unit, and the seller's going-in cap rate and the NOI it is calculated on.
2. In-place income
The NOI the seller presents, and whether it is in-place, adjusted, pro forma or "stabilised".
3. Occupancy and tenancy
Leased versus occupied versus paying, physical and economic occupancy, largest tenants, and the share of income from the top tenants.
4. Lease profile
WALT, expiries by year, options to renew or terminate, and rent steps.
5. Growth story
The seller's rent growth, mark-to-market and lease-up assumptions.
6. Capital and condition
Recent capex, deferred maintenance, and anything the OM says about roofs, HVAC or structure.
7. Debt and structure
Assumable debt, ground leases, tax abatements and special assessments.
8. What the OM does not say
The gaps: missing lease amendments, an expiring anchor lease mentioned only in a footnote, or a T-12 that ends months before the OM date.

A prompt that works with most AI tools: "Summarise this offering memorandum for an acquisitions team. For each of the eight headings below, give the seller's figure or statement, the page it appears on, and whether the OM presents it as in-place, adjusted or projected. List anything the OM does not cover." Then take the output into the check below.

The OM is the seller's story: check it against the data room

The most useful step in OM review is the one most AI tool comparisons skip: checking the seller's headline figures against the source documents in the room. You can run this check with or without AI. It takes the form of a simple table:

  • In-place NOI

    Where to verify it in the data room
    T-12 operating statement, rent roll, leases
    What to look for
    Does annualised recent T-12 income support rent roll rent? Has the OM added back one-off expenses, assumed free rent has burned off, or left out a market management fee, insurance at current premiums, or tax reassessment on sale?
  • Occupancy

    Where to verify it in the data room
    Rent roll, T-12 gross potential rent and vacancy lines, leases
    What to look for
    Is the OM figure leased or occupied: does it count leases signed but not yet started, or tenants who have moved out but still pay? Physical occupancy from the rent roll; economic occupancy only if gross potential rent for all space, vacant space included, is given, otherwise mark it not calculable. Tenants in occupancy but not paying?
  • WALT and expiries

    Where to verify it in the data room
    Leases and amendments, rent roll
    What to look for
    Expiry dates in the rent roll versus the amendment that last changed each lease's term; whether WALT is weighted by rent or area, and runs to expiry or to the first termination option
  • Rent steps and growth

    Where to verify it in the data room
    Leases and amendments
    What to look for
    Contracted steps versus the OM's growth assumptions
  • Tenant credit

    Where to verify it in the data room
    Leases, guaranties, AR ageing, and estoppels once delivered after PSA signing
    What to look for
    Past-due rent in the AR ageing; guarantor named in the lease; estoppel exceptions
  • Recoveries

    Where to verify it in the data room
    Leases, T-12 recovery lines
    What to look for
    Net, gross or modified gross leases versus how recoveries appear in the T-12
  • Capex and condition

    Where to verify it in the data room
    Property condition report, capex history
    What to look for
    Items the OM omits or describes as "complete"
  • Title and legal

    Where to verify it in the data room
    Title commitment, PSA, recorded memoranda
    What to look for
    Recorded lease memoranda or easements that change the picture

The five-minute reconciliation test

Compare the rent roll's annualised in-place base rent with the T-12's base rent line (excluding recoveries and other income) for the latest month or quarter, annualised: the rent roll is a snapshot on one date and the T-12 averages the past year. Check first whether the T-12 is on a cash or accrual basis: an accrual T-12 with straight-lined rent spreads free rent across the lease term, which hides it. For example, if the rent roll says the occupied space should earn noticeably more than the T-12 shows it earning, look for free rent or concessions, past-due rent or bad debt, an amendment that changed rent, leases signed but not yet paying, or a stale rent roll.

Keep occupancy separate. The rent roll gives physical occupancy. Economic occupancy needs gross potential rent: what all the space, vacant space included, would earn at its asking or market rent over the same period. If neither the T-12 nor the rent roll gives potential rent for the vacant space, record economic occupancy as not calculable rather than estimating it. Collecting all of the occupied space's rent is a 100% collection rate, not 100% economic occupancy.

Ask any AI tool to run this test on a sample pack. Judge the answer on whether it cites both figures, allows for the timing difference, uses the right denominator for any occupancy figure, and says what it could not find. A tool that summarises each document separately will miss the gap.

Leases and amendments matter most

When an OM and the T-12 disagree, the usual reason is that the OM presents adjusted or pro forma NOI. When a rent roll disagrees with the leases, a frequent reason is an amendment that changed rent or term and never reached the rent roll. Read each lease with all its amendments. Each amendment changes only the terms it addresses, so the current rent may come from an earlier document than the latest one.

AI due diligence for a real estate data room: what it should do

Once you move past the OM, diligence splits into six workstreams. Here is what AI should help with in each, and where a person still decides. Terms here follow US practice; the documents and their names differ in other markets.

  • Leases

    What AI can help with
    Abstract key terms from each lease and its amendments; list options, expiries, co-tenancy, exclusives and tenant purchase rights (first refusal or first offer); flag terms that differ from the rent roll
    Where a person still decides
    Whether a non-standard clause is acceptable; what to ask in estoppels
  • Financials

    What AI can help with
    Tie the T-12 to the rent roll and leases; show recoveries and one-off items; draft a first-pass variance explanation
    Where a person still decides
    Underwriting assumptions and the price
  • Legal

    What AI can help with
    Summarise the PSA; list deadlines and conditions, including deposit and due diligence deadlines and estoppel and SNDA delivery conditions; pull out reps, warranties and remedies
    Where a person still decides
    Negotiating position; legal sign-off by counsel
  • Title and zoning

    What AI can help with
    Summarise the title commitment; list exceptions and recorded documents
    Where a person still decides
    Whether an exception is acceptable; survey and zoning advice
  • Physical and environmental

    What AI can help with
    Summarise the property condition report and Phase I environmental site assessment; list recommended repairs and recognized environmental conditions
    Where a person still decides
    Further testing, capex budget, whether to proceed
  • Market

    What AI can help with
    Compare the OM's rent and occupancy assumptions with your own market data and past deals
    Where a person still decides
    The view on the market

What ties these together is evidence. Every figure in the diligence summary should point to the document and page it came from, so the reviewer can check it quickly and the investment committee can see how the conclusion was reached.

Can you upload data room documents to ChatGPT or Claude?

This is the question most teams ask first, and the answer depends more on your paperwork than on the tool. What follows is general information, not legal advice; your NDA and your counsel govern what you may do with deal materials.

Read the NDA
Confidentiality agreements typically define who counts as a "representative" and limit use of the materials to evaluating the transaction. Uploading files to an outside service may or may not be permitted, and an AI vendor is not usually a named representative. Check the return-or-destroy clause too: files uploaded elsewhere are copies you may need to delete when the process ends. The data room's own terms of use may also restrict downloads. If in doubt, ask counsel or the broker.
Use a firm-managed account
Business and enterprise tiers of general assistants offer more control over data use than personal accounts. A personal account on deal documents is the scenario most firms want to stop.
Keep the files where the deal lives
The data room belongs to the seller. Your analysis, notes and IC memo belong to your firm and should live in your firm's own deal site, such as a SharePoint site with permissions for the deal team, so the evidence survives to IC and beyond.
Check what the room records
Data rooms record buyer activity. Find out what the room logs before asking its built-in AI questions that reveal your thesis.
Expect general assistants to stop short
They are good at summarising a narrative document, and with connectors they can reach some of your systems, but they do not keep a review trail behind each figure across dozens of files, which makes verification harder.

For more on that trade-off, see ChatGPT or Claude vs a CRE-built AI platform. For the security questions to put to any AI vendor before deal documents go near it, see the AI vendor security checklist for real estate CFOs.

Template: OM and data room summary

Copy this into a document or spreadsheet for each deal. The second part is the one that matters most.

1. Deal snapshot

  • Asset, address, asset type, size
  • Seller, broker, process stage, bid date
  • Guidance price, price per square foot or unit, seller's going-in cap rate
  • Data room access date and the date of the latest documents

2. Seller claims versus source documents

  • In-place NOI (p. __)

    Seller figure
    [figure]
    Source document checked (page)
    T-12 and rent roll (p. __)
    Source figure
    [figure]
    Status
    Confirmed / Differs / Missing
    Note
    [note]
  • Occupancy (p. __)

    Seller figure
    [figure]
    Source document checked (page)
    Rent roll (p. __)
    Source figure
    [figure]
    Status
    Confirmed / Differs / Missing
    Note
    [note]
  • WALT (p. __)

    Seller figure
    [figure]
    Source document checked (page)
    Leases and amendments
    Source figure
    [figure]
    Status
    Confirmed / Differs / Missing
    Note
    [note]
  • Largest tenant rent (p. __)

    Seller figure
    [figure]
    Source document checked (page)
    Lease and the amendment that last changed rent (p. __)
    Source figure
    [figure]
    Status
    Confirmed / Differs / Missing
    Note
    [note]
  • Recoveries (p. __)

    Seller figure
    [figure]
    Source document checked (page)
    Leases, T-12 (p. __)
    Source figure
    [figure]
    Status
    Confirmed / Differs / Missing
    Note
    [note]

3. Rent roll and T-12 reconciliation

  • Rent roll base rent versus tenant ledgers and the T-12 base rent line
  • Physical occupancy from the rent roll; economic occupancy only where gross potential rent for all space is available, otherwise marked not calculable
  • Differences explained, each with its source

4. Lease and tenant risks

  • Expiries and termination options in the hold period
  • Co-tenancy, exclusives and termination rights
  • Tenant credit: past-due rent, guarantors, rating changes

5. Open diligence requests

  • Missing documents (for example amendments, service contracts, and estoppels once the PSA is signed)
  • Owner and due date for each

6. Questions for the broker

  • One question per "Differs" or "Missing" status in part 2

7. Sources

  • Every document used, with its date and data room path

How to compare AI tools for OM and data room review

Use this checklist on a sample deal pack before you commit to any tool:

Where do documents go?
Does the tool keep files inside your firm's environment, or a vendor environment with clear data terms? Is customer data used to train models?
Citations
Does every figure in a summary point to the document and page behind it?
Cross-document checks
Can it run the five-minute reconciliation test and explain the gap, rather than summarising each file separately?
Amendments
Does it read each lease with its amendments and use the current terms?
Missing documents
Does it say when something it needs is not in the room, instead of guessing?
Permissions
Can access be limited to the deal team, with single sign-on and role-based access?
Audit trail
Can you see who asked what, and what changed, after the fact?
Outputs
Does it produce the formats your IC uses (memo, model, presentation), with the sources kept?
File types
Does it read Excel rent rolls, multi-tab T-12s and scanned leases, not just text PDFs?
Integrations
Does it connect to where your firm already stores documents, such as SharePoint?
Review
Who reviews the output, and how quickly can they check each figure against its source?

Some vendors quote accuracy percentages. Rather than compare numbers measured on someone else's documents, run your own sample pack through each tool and score how quickly a reviewer can verify each figure against its source. That includes us: Fifth Dimension's own claims are below, and it should be tested against this checklist like any other tool.

Where Fifth Dimension fits

Fifth Dimension is AI for commercial real estate, from underwriting to lease abstraction. For deal teams, it connects your deal pipeline, data rooms, models and market intelligence, so decision-ready analysis reaches investment committee with the sources and assumptions intact. See Fifth Dimension for investment managers. Across the platform:

  • Workspaces. A shared place for the context, activity, review and outputs behind an investment decision, synced with your underlying data rooms.
  • Ask across deal documents. Natural language queries across deal documents, with every answer grounded in your own systems and every source cited.
  • Spot inconsistencies. Information that normally sits apart is connected, so changes, inconsistencies and emerging risks are easier to spot.
  • Evidence. Evidence & Controls keeps sources, assumptions, review steps and changes visible so your team can understand how an output was produced.
  • Decision materials. Analysis becomes the memos, reports, models and presentations your team already uses, with the working behind them intact. Every output is built for scrutiny, with source documents cited and reasoning transparent.
  • Security. SOC 2 Type II compliant, ISO 27001 certified and GDPR compliant; customer data is never used to train the models that power Fifth Dimension; each customer operates in a separate product environment; SAML single sign-on and role-based access. See Fifth Dimension security.

Fifth Dimension's AI agent understands your terminology, processes and priorities, and applies them consistently across each task.

Together, we identify the manual processes that are slowing you down across the full lifecycle of how you invest in and manage your assets. Start with the people and work that matter most, then expand as Fifth Dimension connects more of your organisation. Fifth Dimension takes on the work between receiving information and making a decision, while keeping your team in control of the judgement that matters. Request a demo.

FREQUENTLY ASKED QUESTIONS.

What's the best AI tool for summarizing commercial real estate offering memorandums and data rooms?
It depends on the job. A data room's built-in AI, where it has one, helps you search inside the seller's room; a general assistant can summarise documents you are allowed to share, but cross-checks depend on how each analyst prompts it; and a CRE platform fits when the summary must tie back to every source document, stay under your firm's control and reach investment committee.
What is the difference between an offering memorandum and a data room?
The offering memorandum is the broker's marketing document: the seller's summary of the asset and its story. The data room holds the source documents behind it, such as the rent roll, T-12, leases, title and the PSA; tenant estoppels follow once the PSA is signed. A buyer summarises the OM, then checks it against the room.
What is the five-minute reconciliation test?
Compare the rent roll's annualised in-place base rent with the T-12's recent base rent line, annualised, since the rent roll is a snapshot and the T-12 is trailing. Leave out recoveries and other income, and check whether the T-12 is on a cash or accrual basis: straight-lined rent hides free rent. If the rent roll implies noticeably more rent than the T-12 shows, look for free rent, past-due rent, an amendment that changed rent, leases not yet paying or a stale rent roll. Keep occupancy separate: economic occupancy needs gross potential rent for all space, vacant space included.
What should an OM and data room summary include?
A deal snapshot, a table of seller claims versus source documents with a status for each, a rent roll and T-12 reconciliation, lease and tenant risks, open diligence requests, questions for the broker, and a list of every source used.
What is a T-12, and why does it matter in OM review?
A T-12 is the property's trailing 12-month operating statement: income and expenses for the last twelve months. It shows what the asset actually earned, so it is the main check on the in-place NOI and rental income the OM presents.
How does Fifth Dimension help with OM and data room review?
Fifth Dimension gives deal teams a shared Workspace synced with their underlying data rooms, natural language queries across deal documents with every source cited, and Evidence & Controls that keep sources, assumptions, review steps and changes visible. It is SOC 2 Type II compliant and ISO 27001 certified.

Tools & libraries.