Screen Every Inbound Deal Automatically: The CRE Deal-Inbox Playbook | NextAutomation

To screen every inbound deal automatically, you stop letting brokers set the format. Instead of a raw PDF landing in your inbox and sitting there until someone has a spare hour, every inbound deal enters through one dedicated address, gets extracted into a structured record, passes a completeness gate so nothing enters half-documented, and is then ranked against the criteria for the specific mandate it is competing for. AI does the capture, the extraction, the completeness check, the scoring, and the first-draft memo. An analyst reviews every score before it means anything. That last part is not a hedge; it is the design.

This is written for the acquisitions lead who is drowning in broker emails and offering memorandums. You know the problem: the deals arrive faster than anyone can read them, they arrive in twelve different layouts, and the good one is buried three days deep in an inbox behind the ten that were never going to clear your box. The playbook below is the deal-intake pattern we already run in production, decomposed into five steps you can reason about. It is grounded, not aspirational, and it is honest about the one step that stays human.

Step 1: Capture Every Inbound Deal at a Single Address

The first failure of the manual deal desk is not analysis. It is capture. A broker sends an offering memorandum, a flyer, or a two-line email with the deal buried in an attachment, and it lands in a personal inbox next to a hundred other threads. Nothing is standardized, nothing is logged, and the deal exists only as an unread email until a human happens to open it.

The fix is a single dedicated intake address. You forward the broker email, the OM, or the flyer to that address, and an AI extraction step reads the attachment and fills in a structured deal record: property, location, asset type, unit or square-foot count, price, and the financial fields it can find. What was an unread PDF becomes a row you can actually work with. Critically, this is capture, not a decision. A human still reviews what was extracted, because the point of Step 1 is only to get the deal out of the inbox and into a structured shape where the rest of the process can run.

That last sentence matters. The extraction is fast and it is consistent, but it is not infallible, which is exactly why the next step exists. The value of Step 1 is not that AI perfectly reads every document on the first pass. It is that every inbound deal now enters through one door, in one format, logged, instead of scattering across inboxes where deals quietly die of neglect.

Step 2: Run the Completeness Gate Before Anything Else

Once a deal is a structured record, the next question is not "is this a good deal." It is "do we even have enough to judge it." Half the deals that reach an acquisitions desk are missing something material: no rent roll, no trailing twelve, no expense detail, a price with no basis behind it. Scoring an incomplete deal produces a confident number built on holes.

So before any scoring happens, every inbound deal runs through a completeness gate. In the pattern we run, that gate is a 26-point checklist applied to every record. It does one job: it flags which required fields are missing so nothing enters the pipeline half-documented. If the OM never included the trailing twelve, the checklist says so. If the rent roll was referenced but not attached, the checklist catches it. This is also the backstop for Step 1: if the extraction missed a field the document did contain, the completeness gate surfaces it as missing, and a human can go pull it.

Be clear about what this step is and is not. The 26-point checklist is a completeness gate, not a decision step. It never says a deal is good or bad. It says a deal is documented or it is not, and it tells you exactly what to chase from the broker before you waste an analyst's afternoon underwriting a hole. Getting the go or no-go right depends on first knowing you are judging a complete picture, and this is the step that guarantees it.

Step 3: Apply Your Disqualifiers and Weighted Criteria, Per Mandate

Now the deal is captured, structured, and documented. Only now does screening in the real sense begin, and it happens in two moves: cut the dead deals, then rank what survives.

First, the automatic disqualifiers. In the pattern we run, three automatic disqualifiers remove deals that fail a hard constraint outright. These are the bright lines a mandate cannot cross: an asset type it does not touch, a market it does not buy in, a size band it will not consider. If a deal trips one of these, it is out, and no analyst time is spent on it. That is the single biggest source of wasted screening effort recovered, because a huge share of broker inbound is simply not a fit and everyone knows it after ten seconds of reading; the disqualifiers just make those ten seconds automatic.

Second, the deals that clear the disqualifiers get ranked against five weighted criteria. These are the factors that actually separate a deal worth an analyst's full attention from one that is merely acceptable, weighted according to what this mandate cares about most. The output is an ordered list: the deals most worth your time at the top, the marginal ones below.

The word that makes this work is per mandate. The three disqualifiers and five weighted criteria are not a universal scoring engine handed down from a vendor. They are configured for the specific fund, strategy, or buy box the deal is competing for. A value-add multifamily mandate and a stabilized industrial mandate disqualify different deals and weight different factors, and the same inbound deal can rank high for one and get cut by another. Configuring these to your actual mandate is the difference between a screen that reflects how you invest and a generic filter that fights you.

Step 4: Keep an Analyst on Every Score (the Honest Step)

Here is the step that most "AI deal screening" pitches quietly skip, and it is the one that makes the rest trustworthy. The AI screens, scores, and drafts. It does not make the investment decision. An analyst reviews every score and every memo before it counts.

Say plainly what the AI is and is not doing. It extracts the record, runs the completeness gate, applies the disqualifiers, ranks against the weighted criteria, and drafts a first-pass screening summary. Every one of those is a mechanical or structured task that a machine does faster and more consistently than a person reading PDFs at 6pm. What it is not doing is deciding to pursue a deal, price a deal, or take a deal to committee on its own authority. A person owns that, always.

This is not the AI hedging its bets. It is where the actual leverage lives. The value of the whole system is not that a machine replaces your judgment. It is that your analyst starts every deal from a structured, documented, pre-ranked record instead of a raw PDF and a cold start. They open a deal already knowing what fields exist, what is missing, whether it cleared the hard constraints, and where it ranked, and they spend their scarce judgment on the deals that earned it. Any product that tells you the AI decides is describing a liability, not a feature. The honest version keeps a human on every score, and it is faster anyway, because the human is reviewing instead of transcribing.

Step 5: Wire It Into Your Actual Pipeline

A screening engine is only useful if it lives where your team already works. This is the step acquisitions leads ask about most, usually in the form of very practical questions, so here are the honest answers.

Where does a screened deal show up? In the deal record itself, in the pipeline view you already use to track deals through their stages. A screened deal is not a report that lands in a separate tool you have to remember to check. It appears as a record in your pipeline with its extracted fields filled, its completeness flags visible, its disqualifier status set, and its rank assigned, sitting in the stage it belongs in. You work it where you work everything else.

And when documents are missing from a broker, where does that show up for you? This is a real question we get asked, almost word for word: when a document is missing from a deal, where does it show for me, in my pipeline tool? The answer is yes, right on the deal record. The completeness gate from Step 2 writes its missing-field flags onto the record, so the same view that shows you the deal shows you exactly what is still outstanding. You are not hunting through the original email to figure out what the broker left out. The record tells you, and you chase it from there.

The point of Step 5 is that none of the previous four steps require your team to change how they operate. The deals still arrive from brokers. Your analysts still make the calls. What changes is that between the broker and the analyst there is now a capture, gate, and rank layer that turns a chaotic inbox into an ordered, documented pipeline. If you want to see how the same discipline extends into producing the committee document itself, we decompose that in our guide to turning an OM into an IC memo with AI. The upstream extraction work, pulling the numbers out of the trailing twelve and rent roll cleanly, is covered in how to extract a T12 and rent roll with AI.

Where AI Fits in the Broader Underwriting Stack

Screening is the front door, not the whole house. Once a deal clears the screen and an analyst decides it is worth real work, it moves into underwriting proper: the model, the assumptions, the sensitivity, the memo. Deal screening earns its keep by making sure only the deals worth that effort ever reach it, and by handing the analyst a structured record instead of a raw document when they get there.

If you are evaluating where to put AI across the full workflow rather than just the inbox, it is worth seeing how the pieces fit. We break down the tooling landscape in our guide to the best AI tools for CRE underwriting, and where screening sits in the broader adoption picture is covered in the state of AI in commercial real estate in 2026. The consistent thread, in screening and everywhere else, is that AI removes the mechanical volume and a human keeps the judgment. A deal-inbox screen that respects that line is one you can actually trust to run every day.

Frequently Asked Questions

When documents are missing from a deal, where does that show for me?

Right on the deal record, in the same pipeline view you already use. A completeness gate runs on every inbound deal and flags which required fields are missing, and those flags are written onto the record itself. So when a broker sends a deal without a rent roll or a trailing twelve, you do not have to go back and re-read the original email to notice. The record shows you exactly what is outstanding, and you chase it from there. That is the whole point of the completeness gate: nothing enters the pipeline half-documented and silently, and you always know what is missing.

How does AI deal screening actually decide which deals are good?

It does not decide, and that is deliberate. The AI applies your automatic disqualifiers to cut deals that fail a hard constraint, then ranks the survivors against your weighted criteria, and drafts a first-pass summary. That produces an ordered list and a starting point, not a verdict. An analyst reviews every score and every memo before it means anything. The AI screens, scores, and drafts; a human decides. The value is that your analyst starts from a structured, pre-ranked record instead of a raw PDF, so their judgment goes to the deals that earned it rather than to transcription.

Do I have to change my tools or how my team works to use this?

No. Deals still arrive from brokers the way they always have, and your analysts still make the calls. What sits between them is a capture, completeness, and ranking layer. You forward the broker email or OM to a single dedicated intake address, and the screened deal shows up as a record in the pipeline view you already use, with its fields extracted, its missing documents flagged, its disqualifier status set, and its rank assigned. The screening happens between the broker and the analyst, without asking your team to adopt a new way of working.

Is the scoring the same for every fund, or is it configured per mandate?

Per mandate, always. The automatic disqualifiers and the weighted criteria are configured for the specific fund, strategy, or buy box a deal is competing for, not handed down as a universal scoring engine. A value-add multifamily mandate and a stabilized industrial mandate disqualify different deals and weight different factors, and the same inbound deal can rank high for one and get cut by another. Configuring the screen to how you actually invest is the difference between a filter that reflects your strategy and a generic one that fights you.

What happens if the AI extracts a field wrong from the OM?

Two things catch it. First, capture is not a decision step: a human reviews the extracted record, so an obvious misread gets corrected before the deal moves. Second, the completeness gate runs on every record and flags missing or absent required fields, which surfaces cases where a field the document contained did not make it into the record. The system is built on the assumption that AI can miss a field, not on the pretense that it never will, which is exactly why the completeness gate and the analyst review exist. That honesty is the difference between a screen you can trust and one that fails quietly.