Tips & Best Practices

How to get the most out of DealBrief.

If you've never used an AI deal tool before, start here. Six minutes of reading saves an hour of redo's.

What's in this guide

  1. How DealBrief actually works
  2. Reading the VERIFIED / ASSUMED / MISSING tags
  3. OM Generator — best practices
  4. Deal Intelligence — best practices
  5. Drag-and-drop files — what works, what doesn't
  6. Reviewing the output before you send it
  7. Common beginner mistakes
  8. When to use OM vs Deal Intelligence vs both
01 — Foundation

How DealBrief actually works.

It's a draft engine, not an oracle. Treat it like a sharp junior analyst who never forgets to flag what's missing.

Behind both tools is Anthropic's Claude AI — the same engine institutional research desks use. DealBrief gives Claude a strict five-step pipeline:

  1. Ingest — Reads everything you provide: form fields, drag-and-drop documents, your notes.
  2. Tag — Marks every fact as VERIFIED (you provided it), ASSUMED (Claude inferred a reasonable default), or MISSING (no data, do not guess).
  3. Reconcile — Checks the math. If you say cap rate is 6.2% but the NOI ÷ price is 5.0%, that gets flagged.
  4. Draft — Writes the report following institutional CRE structure (Executive Summary, Financials, Risks, etc.).
  5. Close — Generates a Diligence Checklist of every MISSING item you need to source before this document leaves your hands.

What this means for you

The output is only as good as your input. Give it good data, you get a sharp draft. Give it sparse data, you get a draft with lots of MISSING tags — which is still useful because it tells you exactly what to source before sharing the document.

02 — The most important concept

Reading the three tags.

Every figure, every claim, every line item in your output is tagged. Learn these three and you can trust the output.

VERIFIED

You provided this directly. The AI did not invent it. This is your responsibility — if the number is wrong, it's wrong because you put a wrong number in.

ASSUMED

The AI inferred this with the reasoning shown. Example: you gave price + total units → AI calculates price per unit and tags it ASSUMED with the formula. Treat as a working estimate.

MISSING

No data was provided and the AI refused to invent one. This is a strength, not a weakness. Source the data, fill the field, re-run the report.

Why this matters

Most AI tools make stuff up. DealBrief is built to refuse. If a buyer or principal asks "where did you get this number?" — you can point to VERIFIED (you provided), ASSUMED (with shown reasoning), or MISSING (you sourced it post-draft).

03 — Offering Memorandum

OM Generator best practices.

An OM is a marketing document for a deal. The goal is to position the asset clearly, defensibly, and within institutional standards.

The 5 fields that matter most

If you fill out only five fields, fill these:

Everything else is upside. If you have it, paste it. If you don't, leave it blank — the AI will mark it MISSING and put it on your checklist.

Writing strong "Investment Highlights" notes

The "Investment Highlights" field is where you bias the narrative. Be specific.

Weak input

Good location, value-add, rent upside.

Vague. AI has nothing to amplify.

Strong input

3 blocks from Brickell City Centre; 56 of 84 units unrenovated with $200/mo rent lift demonstrated on prior 28 renovated units; $20M assumable loan at 4.85% fixed through 2028.

Specific. AI turns each fact into a defensible bullet.

Always fill the "Known Risks" field

The OM will include a Risks section whether you fill this or not. If you fill it, the AI gives those risks the right framing (paired with mitigations). If you don't, the AI generates generic risks. Your version is always better.

Pro tip — How to handle a sparse OM

You can run an OM with just Property Type, Address, and Asking Price. The output will be ~60% MISSING tags. That's not a failed run — that's a perfect intake checklist. Use the Broker Pre-Distribution Checklist at the bottom as your data-gathering punch list, then re-run.

What the OM Generator is NOT for

04 — Deal Intelligence

Deal Intelligence best practices.

Where OM Generator is a marketing draft, Deal Intelligence is an analytical read — the "should we pursue?" sharp-pencil view.

The 6 fields that matter most

Understanding the "Analytical Position"

Every Deal Intelligence report ends with one of three positions:

Pursue

Numbers, comps, and structure align with your return target. Move to LOI.

Pursue with Conditions

Deal works under base case but several items must be verified first. The conditions are listed — work through them.

Pass

Either the numbers don't work at this price, or the risks materially exceed the return. Either pass or negotiate price down.

This is not GO/NO GO advice

The Analytical Position is one analyst's read on the data you provided. It's not a recommendation. An independent appraisal and your own underwriting judgment are required for any binding offer. The report says this explicitly in every output.

How to use the Diligence Checklist

The Diligence Checklist at the bottom of every Deal Intelligence report is your pre-LOI sourcing list. Hand it to your analyst, your title company, or your environmental engineer. Each item is something specific to verify before binding.

Standard institutional buyers will check 90%+ of these. If you can hand them a report that already lists what they'll ask for, you look prepared.

05 — Drag and drop

Drag-and-drop files.

Both forms accept files. The AI reads them and pre-fills the form. This is the fastest way to run a deal.

What works well

What needs converting

Pro tip — Drop multiple files at once

You can drop a rent roll, a T-12, and a property summary together. The AI reads all three and reconciles fields across them. If the rent roll shows $1.92M annualized but the T-12 NOI says $1.85M, you'll see both numbers tagged with their source.

Privacy reminder

Files are sent to Anthropic's Claude API for processing. We never store the file content on our servers. You are responsible for redacting any personally identifiable information (SSNs, bank account numbers, personal addresses) before uploading. Most CRE files don't contain PII, but rent rolls sometimes have tenant names — your call.

06 — Before you send

Reviewing the output before distribution.

DealBrief outputs are drafts. Treat the AI like a junior analyst — review their work before forwarding.

The 5-minute review checklist

Marketing PDF vs Working PDF

Each report can be downloaded as two PDFs:

Marketing PDF

Clean version for sharing with buyers, principals, lenders. All VERIFIED / ASSUMED / MISSING tags are stripped. Requires you to check a box certifying you've reviewed every figure. Use this for external distribution.

Working PDF

Internal version with all tags preserved. Shows you exactly what's verified, what's assumed, what's missing. Use this for your own analysis, your team, your file.

Pro tip — Always download the Working PDF first

Even if you plan to send the Marketing PDF, download the Working PDF too. It's your audit trail of what the AI saw vs. what it inferred. Keep it in your deal file.

07 — Don't do this

Common beginner mistakes.

1. Treating the output as final

The AI-Generated Draft Notice on every report is not boilerplate — it's a working rule. Every figure must be broker-verified before distribution. Don't skip this.

2. Filling the "Risks" field with "none"

Every deal has risks. If you leave Risks blank, the AI generates generic ones (insurance, vacancy, etc.). Your version — specific to this deal — is always sharper. Fill it.

3. Inflating numbers to make the deal look better

The AI tags broker-provided figures as VERIFIED. If you provide a wrong NOI to make the cap rate look better, the report will defend that number — and your reputation rides on it when a buyer's diligence team checks. Use real numbers.

4. Using OM Generator when you needed Deal Intelligence

OM Generator = marketing draft (sell the deal). Deal Intelligence = analytical read (should we pursue?). If you're underwriting an acquisition, run Deal Intelligence. If you're listing a property, run OM Generator. Sometimes you need both — see next section.

5. Stripping the AI-Generated Draft Notice

The disclaimer at the bottom of every output is your legal cover. It's small. Don't remove it. The Marketing PDF keeps it as a footer for exactly this reason.

6. Running once and giving up

First run on a deal with sparse data will have lots of MISSING tags. That's fine. Use the checklist, source the data, re-run. Your second pass will be 80% better. Free tier gives you 3 runs per product — that's enough for one deal with two refinements.

08 — Choosing the tool

OM, Deal Intel, or both?

Use OM Generator when…

  • You're listing a property and need a marketing draft
  • You're sending an Offering Memorandum to potential buyers
  • You need a structured deal narrative for a CA-protected blast
  • You're packaging a deal for a broker's tour

Use Deal Intelligence when…

  • You're considering a deal as a buyer or principal
  • You need to stress-test someone else's pitch
  • You want to see seller and buyer perspectives side by side
  • You're prepping for an investment committee

Use both when…

You're listing a property and want to pre-anticipate what a sharp buyer will push back on. Run Deal Intelligence on your own listing — read it as the buyer would — then strengthen the OM accordingly. This is the move that separates listing brokers who close from those who lose deals in diligence.

Ready to try it?

You have 3 free OMs and 3 free Deal Intelligence reports. Start with whichever fits the deal in front of you.