Private equity deal screening that tests every CIM against your fund's own limits, with the page that decided each gate
Every CIM tested against the fund’s own limits in minutes, with every page cited.
See one case, screen by screen ↓Private equity deal screening is the first pass on every CIM and teaser a fund receives: testing it against the fund's investment criteria, checking the numbers and deciding to decline or take a first look. Private Equity Deal Screening is deal screening AI that does that pass with seven agents: they check the information barrier, test the LPA, side letters and IC guidelines gate by gate, challenge the add-backs and draft a cited one-page screen, while the associate decides and a VP confirms.
AI in asset management: a buyer's guide to agentic AI for investment managers →
Updated
Private equity deal screening is the first pass on every CIM and teaser a banker sends. The confidential information memorandum review alone means reading 60 to 120 pages; then the associate tests the deal against the fund's LPA investment restrictions, side letters and IC guidelines, rebuilds management's adjusted EBITDA, checks whether the firm has looked at the company before, writes a one-page screen for the VP, and either declines politely or asks for a first look. A private equity team at a large manager receives more than a thousand of them a year and closes a handful, so the first CIM review is skimmed under deadline and the limits are checked from memory.
At a multi-asset manager there is a step before all of that. The CIM is material non-public information, and the firm's public bond and equity funds may hold the target's securities. The control room often hears about it days later, if at all. Meanwhile management's add-backs are taken at face value, the same company comes back under a new code name from a new banker, and a decline goes into the deal CRM as "pass", with no reason code and no page an examiner could follow.
Estimated split for the first pass on a mid-size buyout CIM, by hand and with the solution, including the associate's review.
Seven agents check the barrier, test the mandate, challenge the numbers and draft the screen; the deal team decides and signs.
Monday, October 19, 2026, New York. At 9:11 a.m. Harcourt Lane Advisors sends Project Atlas: an 84-page CIM for Anvara Industrial Group B.V., a Rotterdam maker of flow-control valves and actuators, with its Turkish and Dutch statutory accounts. IOIs are due Friday, October 30. Here is what happens next, screen by screen, in the working solution.
Nina's Today page shows three numbers: 7 deals in her queue, 3 IOI deadlines in 14 days (Pine, Atlas, Birch) and 1 decline awaiting a VP. Project Atlas has just arrived from Oliver Pratt: 84 pages in English, with Turkish and Dutch filings. Below it, what needs her: Birch ready to advance at a score of 86, Spruce needing her judgment on leverage, Ember sent as a password-protected file.
From your agents: “Held Project Mercury unopened: its listed parent is held by Quillfield’s public equity funds.”
The screen opens empty and the agents work in a conversation beside it. The Intake agent reads all three documents and records the target, the banker, the IOI date and NDA-2026-1187 on file. Each gate in the row waits until the barrier check is done, then lights in turn.
“The agents are screening. You decide once the screen is drafted. Nothing is sent or decided without you.”
Anvara has 6.125% senior secured notes due 2029, and three Quillfield public portfolios hold $41.2M face: the Global High Yield Fund $22.6M, the European High Yield UCITS Fund $13.9M and a public pension's separate account $4.7M. By standing rule SR-2 the Wall agent puts the issuer on the watch list (WL-2026-0918), limits Atlas to six named people and blocks three public-side teams.
“By Standing rule SR-2 … I put the issuer on the watch list and limited Project Atlas to six people. Ruth, please confirm.”
Sector passes (industrial technology, CIM p. 9). Size passes: enterprise value guidance $640M–$700M, an equity check of about $300M, 4.4% of the fund. Reported EBITDA margin is 14.2%. Leverage is 5.76x on Quillfield's EBITDA, inside 6.0x. Exclusions pass with a note: naval valves are 4.1% of revenue, conventional. Geography fails. Score 58 of 100: recommend decline.
Every gate shows the rule, what was found, pass or fail, and the page that decided it.
The page that decided it is CIM p. 23: Turkey $169.2M of $412.6M LTM revenue. LPA §4.2(c) allows no single country outside North America and Western Europe above 25%. This is an LPA restriction, not an IC guideline, so a partner cannot override it; only the fund's LPAC can waive it, and the screen offers to request that consent rather than assume it.
“An LPA restriction: only the fund’s LPAC can waive it.”
The Turkish subsidiary's 2025 statutory accounts are read in the original, with the English beside them. Page 12 reports revenue (hasılat) of TRY 6,412.5 million; at the 2025 average of 39.40 TRY/USD that is $162.8 million, the same as the CIM's 2025 Turkey line on page 23. Every figure carries its citation, ready to copy.
Management bridges $58.4M reported EBITDA to $71.9M adjusted with $13.5M of add-backs, 23.1% of reported. The Skeptic accepts the Tilburg restructuring, the family holding's fees and a settled legal dispute; questions the IAS 29 hyperinflation loss and FX losses on euro loans as recurring; and rejects $2.3M of Manisa plant savings planned for Q2 2027 and not yet made. Underwriting EBITDA: $65.5M.
“Accepted $7.1M · questioned $4.1M · rejected $2.3M.”
The Memory agent matches Atlas to Project Meridian, March 2023, on the chamber of commerce number and the Izmir plant address: declined at first look for Turkey at 38% of revenue. The Allocation agent tests the other three funds under policy AP-3, and none fits. Nina asks what a sale without the Turkish business would look like: $243.4M of revenue, and geography would pass.
“The CIM does not split Middle East and Africa by country, and gives no EBITDA for the remainder: I would ask the banker before relying on it.”
Nina picks D-GEO, Outside geography limits, and signs. The drafted note thanks Oliver Pratt for sharing Project Atlas, says the company's geographic profile falls outside the fund's investment parameters and leaves the door open if the perimeter changes. It says nothing learned from the CIM, and it is sent only after a Vice President confirms.
“A scored deal: a Vice President (Marcus Oyelaran) confirms. You cannot confirm your own decline.”
Marcus opens his Sign-offs and sees the Atlas decline with its reason in one line: Turkey 41.0% against the 25% cap, LPA §4.2(c), score 58. He confirms with his signature, or returns it with a reason. The deal CRM records the stage and the reason code, and the watch-list entry stays until the CCO releases it.
“The maker never checks their own work.”
The same CIM, seen by four of the people who carry it — what their week looked like, and what it looks like now.
Seven agents check the barrier, test the mandate, challenge the numbers and draft the screen; the deal team decides and signs.
Collects CIMs, teasers and filings from the deals inbox, data rooms and uploads, in any language; identifies the project, target, banker, IOI date, language and NDA, and flags files it cannot read.
Checks the target, its parent and its securities against the restricted and watch lists and the firm's public holdings, adds the issuer to the watch list by standing rule, builds the deal roster and hides the deal from the public side.
Tests the CIM against every gate of the fund mandate — sector, size, EBITDA margin, geography, leverage and exclusions — with the page that decided each one, and scores the deal.
Rebuilds EBITDA on a second model from the CIM and the target's statutory filings in their original language, classifies every add-back as accept, question or reject with its page, and recomputes leverage.
Finds prior looks, duplicates and relationships across every screen since 2019 and the deal CRM, matched on registry numbers and addresses rather than code names.
Tests every Quillfield fund's mandate and applies the written allocation policy AP-3, and works out what would change the answer.
Writes the one-page screen, the deal screening memo, in the house template, the decline note to the banker and the weekly pipeline note, every figure cited.
Propose, confirm and sign. The agents propose; a named person decides.
Ask the agents about a CIM, a gate, an add-back or a fund — or test what would change the answer. Every answer cites its page, and nothing changes without you.
Why did Project Atlas fail its screen?
Geography fails an LPA restriction: Turkey is 41.0% of revenue ($169.2M of $412.6M, CIM p. 23) against the 25% single-country cap in LPA §4.2(c) of Quillfield Capital Partners V. The Turkish statutory accounts confirm the 2025 figure. Five other gates pass.
What if the Turkish business were carved out?
Without the Turkish business revenue would be $243.4M: Western Europe 46.5%, Middle East and Africa 24.1%, North America 16.6%, Latin America 12.9%. No single country outside North America and Western Europe would exceed 25%, so geography would pass. The CIM gives no EBITDA for the remainder, so I would ask the banker before relying on it.
Show me the add-backs
Management adds back $13.5M (23.1% of reported EBITDA). I accept $7.1M, question $4.1M (the IAS 29 monetary loss and FX on euro loans) and reject $2.3M of Manisa savings not yet made.
Which deals this week fit two funds?
One: Project Birch, a German test-and-measurement maker, fits European Mid-Market III and Capital Partners V. Allocation policy AP-3 §2 points to QEMM III; because two funds fit, Ruth Adeyemi decides before any first look.
13 screens from the working solution, on its sample data. Pick one to see it large.
The thresholds every CIM is screened against are settings. The administrator proposes a change, a second person checks it, and nothing in the settings can decide a deal.
| Setting | Default | Choose from |
|---|---|---|
| Gate confidence threshold | 90% | Below it, a person decides the gate |
| Score bands | ≥ 75 first look · 50–74 judgment · < 50 decline | An LPA fail recommends decline whatever the score |
| Score weights | Mandate fit 40 · financial quality 25 · earnings quality 15 · risk 20 | Changed by the administrator, with a checker |
| VP confirms every decline on a scored deal | On | Maker-checker; the proposer can never confirm |
| Teaser declines may be confirmed in a batch | On | Only for teasers that failed a hard gate |
| LPA restriction can be waived only with LPAC consent | On | Locked: set by the LPA |
| A CIM screened after receipt | Within 1 business day | Escalates to the VP after that |
| Screens, decisions and barrier records | Kept 7 years | Rule 204-2 requires 5; CIM originals returned or destroyed 30 days after a decline, as the NDA requires |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample deal data · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (4–8 hours: typical vendor-published range for a preliminary memo; S&P Global Ratings EBITDA add-back studies, Feb 2023 and Mar 2024 — add-backs 29% of management-projected EBITDA for 2022 deals; SEC Section 204A action against a private equity adviser, Dec 2023, $4 million). People, firms, funds, companies and advisers named on this page are fictional — characters and sample data in the working solution.
Private equity deal screening is the first pass on every CIM and teaser a fund receives: testing it against the fund's investment criteria, checking the numbers and deciding to decline or take a first look. Private Equity Deal Screening is deal screening AI that does that pass with seven agents: they check the information barrier, test the LPA, side letters and IC guidelines gate by gate, challenge the add-backs and draft a cited one-page screen, while the associate decides and a VP confirms.
Each fund's LPA, side letters and IC guidelines are read into gate rules, each citing its clause, checked by the CCO and versioned. The Screener agent tests every CIM against those gates — sector, size, EBITDA margin, geography, leverage and exclusions — and shows the page that decided each one. It tells an LPA restriction, which only LPAC consent can waive, from an IC guideline a partner can override with a reason.
The Skeptic agent runs on a second, independent model and rebuilds EBITDA from the CIM and the target's own statutory filings in their original language. It classifies every add-back as accept, question or reject with its page and recomputes leverage on the firm's number. The associate can overrule any line with a reason, and the totals recompute.
Before anyone opens the CIM, the Wall agent checks the target, its parent and its securities against the restricted and watch lists and the firm's public holdings. If a public portfolio holds them, the issuer goes on the watch list by standing rule, the deal is limited to a named roster and public-side teams are blocked by entitlement. The CCO confirms each entry and decides holds and any wall crossing. Section 204A of the Advisers Act asks for MNPI policies that fit how an adviser actually receives MNPI, and a CIM is one of those ways.
Yes. CIMs, teasers and filings can arrive in any language. On the sample deals the agents read the Turkish and Dutch statutory accounts behind Project Atlas and a German CIM for Project Birch, and show the original page beside the English, with every figure cited to its page.
The Memory agent searches every screen since 2019 and the deal CRM, matching on registry numbers and addresses rather than code names. It finds the earlier look, who decided it and why, and flags the same company sent by two bankers; a person merges the duplicates.
Yes. The agents propose and never decide, send or allocate. The associate declines or advances with a reason code, a Vice President confirms every decline on a scored deal, a deal partner approves every first look, and the CCO decides barrier questions and any deal that fits two funds.
The Agentic Solution Engine builds and deploys it from your requirements and documents — your LPAs, side letters, IC guidelines and a sample of past screens — and it goes live once every quality gate has passed. Setup includes a backtest on past deals, and the CCO approves go-live.
Every KPI cited to its page, every covenant tested the day the pack lands.
1-min demoVirtual Data Room Due DiligenceEvery document read and every red flag cited to its page, before you sign.
1-min demoQuality of Earnings AnalysisA first-cut QoE in three days, every add-back ruled and cited before the price is set.
1-min demoWe'll run Private Equity Deal Screening on a sample of your own past CIMs, tested against your fund's LPA.