From offering memorandum
to IC memo in an afternoon.
Acquisitions runs on capacity, and capacity is the trap: every deal has to be parsed by hand before anyone can judge it, so the team screens what it can reach and waves the rest by. Built AI removes the parsing, drop any document and it extracts, runs your scenarios and drafts a cited IC memo.
| Case | IRR | EM | Yr1 DSCR |
|---|---|---|---|
| Downside | 11.2% | 1.6× | 1.18× |
| Base | 16.4% | 2.1× | 1.31× |
| Upside | 21.8% | 2.7× | 1.44× |
The deals you lose are the ones you never got to underwrite
An acquisitions desk is only ever as fast as its slowest manual step, and that step is not judgment, it is data entry: re-keying a hundred-page memorandum into the model one cell at a time before anyone can form a view.
That is the better part of a day gone before the first scenario is even run. And the data you need to underwrite well is never in one place. The OM gives you the seller's story. The rent roll is a separate file with its own format. The market comps are scattered across a broker's email, a data subscription and last year's deals nobody has tagged. The loan package is a stack of PDFs. To pull a clean view together you assemble all of it manually, and to do it for one deal is a chore. To do it for the dozen that crossed your desk this month is impossible, so most of them get a glance and a pass.
That is the capacity trap, and it is expensive in a way that never shows up on a report. The deals you screen are not necessarily the best deals; they are the ones you had the hours to reach. The rest get a cursory look and a polite decline, and somewhere in that discard pile is the deal you would have chased if you had had the time to see it clearly. Meanwhile a faster firm with the same conviction and a head start on the work is already in exclusivity.
None of this is analysis. Reading an OM to find the cap rate is not insight; it is transcription. Re-keying a rent roll into a model is not underwriting; it is data entry with a deadline. The judgment (is this the right basis, is the business plan credible, does the financing work, what breaks the deal) is the part that deserves your analysts' time, and it is exactly the part that gets compressed because the transcription ate the day. The work is upside down, and capacity is the reason.
Drop the document. Run your scenarios. Draft the memo.
The week-long screening grind becomes a single auditable pass. Extract, simulate, decide. See the engine behind it →
Built AI turns the week-long screening grind into a single auditable pass. Drop any deal document and the Deal Underwriter extracts up to 114 structured fields, each with a pointer to the exact page and line it came from; anything ambiguous is flagged for a human, never guessed.
Those fields hydrate the knowledge graph, and the deterministic calculation engine takes over. This is the part that matters: no number on your memo is the model's guess. IRR and equity multiple across the hold, DSCR and debt yield against the proposed financing, the waterfall distribution across the promote tiers, occupancy and rollover from the rent roll, all of it computed by a real engine, the same way every time, with the work shown cell by cell. Then your scenarios: a rate move, re-leasing at market, an extended lease-up, a softer exit cap, a delayed refinance. Each one is a genuine computation against the same model logic your firm already trusts, not a narrative about what might happen.
And then the platform writes. The output is a branded IC memo on your template: the deal summarized, the basis and the business plan stated, the base case and the sensitivity tables laid out, the risks named, and every figure linked back to the clause, line or assumption it rests on. A document that arrived as an unstructured PDF in the morning is, by mid-afternoon, a committee-ready memo your team reviews and signs, or passes on, with the reasons on the record. The same pipeline is described end to end on the platform page.
An IC memo full of numbers nobody can trace is a liability in the room. Every figure on a Built AI memo resolves to its source, whether a cell in the rent roll, a clause in the loan package or a line in the OM, so when a committee member asks where the in-place NOI came from, the answer is one click to the document, not a scramble back through the model. The memo defends itself.
What the acquisitions team actually gets
Not a faster spreadsheet and not a chatbot in a sidebar. Four changes to how the desk operates, each one giving your analysts back the hours the transcription used to take.
Any document becomes structured data in minutes
Drop a term sheet, OM, loan package or rent roll and up to 114 fields are extracted with a field-level audit trail and confidence flags. The day of reading and re-keying becomes a few minutes of review, and the data lands clean and cited instead of typed and error-prone.
Your scenarios run on the engine, not the model's imagination
IRR, equity multiple, DSCR, debt yield, the waterfall and your sensitivity tables are computed deterministically against your own model logic. Rate moves, re-leasing, lease-up delays and exit-cap shifts are real calculations you can defend, shown cell by cell.
The IC memo is drafted on your template, fully cited
The deal summary, the base case, the sensitivities and the risks are written up in your firm's format with every figure traced to source. You review and sign a near-final draft instead of building one from a blank page at midnight before committee.
The capacity trap breaks, and you screen everything
When parsing stops being the bottleneck, the deals you underwrite are the deals worth underwriting, not just the ones you had the hours to reach. The desk screens more, moves faster on the ones that matter, and stops losing the good ones to firms that were simply quicker to a view.
| The work | Today | With Built AI |
|---|---|---|
| Reading the OM | A day of manual extraction and re-keying | Minutes: up to 114 fields lifted with audit trail |
| Market & comp data | Scattered across email, subscriptions, old deals | Pulled into one queryable graph alongside the deal |
| Underwriting math | Re-built in the model each time, error-prone | Computed deterministically, shown cell by cell |
| Sensitivities | A handful, if there's time before committee | Run in full: rate, re-lease, lease-up, exit cap |
| The screening memo | Days, often built the night before | A branded, cited IC memo in an afternoon |
| Deals you can reach | Only what the team had hours for | Effectively all of them |
It runs on your model and your templates, not a new system to learn
The platform reads your existing model and drafts on your template. Your model, your templates, your data room. See the extraction model →
The fastest way to lose an acquisitions team is to ask it to abandon the underwriting model it has refined over a hundred deals. Built AI does the opposite. The platform reads the structure of your existing model, maps its inputs to graph nodes, and runs scenarios against it, so the model your analyst built and the engine the firm trusts are the same model, not two diverging copies. The memo it drafts comes out on your template, in your format, with your branding.
Your Excel underwriting model
The platform reads your model's logic and maps its inputs to the graph. Extracted deal fields hydrate it, and the engine runs your base case and sensitivities against the same assumptions your team already trusts, with no rebuild and no parallel copy to reconcile.
OMs · term sheets · rent rolls
Every deal document is extracted into the graph with a field-level audit trail. The unstructured stack of PDFs that used to take a day to read becomes queryable, computable and citable, and precedent deals stay searchable alongside the live one.
Your IC memo template
The memo is authored in your firm's branded template and staged for review. Every figure links back to its source, and nothing is sent anywhere. The destination is a place a person sends to, never a place an agent sends from on its own.
And because every deal you run flows into the graph, precedent compounds. The comps from last quarter's underwriting, the assumptions you used, the structures you have seen all stay queryable, so a new deal is read against the backdrop of every deal that came before it rather than from scratch. The full integration and extraction model lives on the platform page.
Carlos Olea, CFO, Howard Hughes Holdings (NYSE: HHH)
Drop a deal. Watch the memo write itself.
Bring a live offering memorandum, term sheet or loan package and watch the platform extract it, run your scenarios on the engine you trust, and draft a branded IC memo with every number cited, in an afternoon, not a week. No migration, nothing leaves your tenant, and nothing acts without your sign-off.