What does AI due diligence actually do?

AI due diligence parses CIMs, tax returns, P&Ls, and customer/supplier data to surface signals a human reviewer might miss in a 30-minute scan: revenue concentration trends, add-back inconsistencies, owner-comp anomalies, and operational dependencies. It does not replace formal QoE or legal review — it screens deals for whether formal diligence is worth the spend.

How is this different from generic ChatGPT analysis?

Acquidex AI is methodology-disclosed: the underlying framework is published, versioned, and aligned with SBA SOP 50 10 8. Generic LLM analysis hallucinates citations and patterns; we cite the specific document section every signal came from. If the AI cannot cite, we do not claim.

Can lenders or CPAs use this?

Yes. Acquidex AI is designed as a screening layer for the deal-chain, not just buyers. Lenders use it to triage acquisition packages before underwriting. CPAs use it to scope formal QoE engagements. The output is the same standardized score every party can reference.

Acquidex · Tool · AI · Pre-LOI Screen

AI Due Diligence · v1.0

Updated 2026-05-09

AI due diligence, with disclosed methodology.

Surface CIM red flags, owner-comp anomalies, and concentration patterns in minutes. The model surfaces the same structural concerns a CPA-led pre-LOI review would, against documented prompts, with explicit confidence reporting per finding.

The Read

Generic LLMs summarize. Acquidex AI underwrites. Every signal cites the specific document section it came from — if the model cannot cite, it does not claim. The output is a single 0–100 score plus an itemized findings ledger every party at the table can reference.

§ 01 · Flow

How the scan runs.

01
Upload acquisition documents
Submit the CIM, three years of tax returns, trailing-12 P&L, and any available customer or supplier concentration data. The AI indexes the document set against the Acquidex underwriting framework.
02
Surface cash-flow signals
The model evaluates add-back density, replacement-labor coverage, and broker-to-tax-return revenue reconciliation, citing the specific document section every signal came from. If the AI cannot cite, it does not claim.
03
Screen pricing and lender readiness
Each deal is band-placed against the relevant Industry Atlas multiples band, then DSCR-pressure-tested at SBA SOP 50 10 8 thresholds. Top-of-band asks without supporting structural conditions are flagged.
04
Map transition risks
Customer concentration, owner-held licenses, lease assignment language, and key-person dependencies are extracted from the CIM and CPA work papers. Each finding ships with the source citation and confidence level.
05
Produce a deal score
A single 0–100 score plus an itemized findings ledger — the same standardized score every party at the table can reference: buyers, lenders, brokers, CPAs.

§ 02 · Coverage

What the model scans before you spend on QoE.

SCANS

Signal areas the AI evaluates

Financial reality

High

Sources · P&Ls · Tax returns · Bank activity

Multi-year financials cross-checked for add-back quality, seasonality, and phantom margin. DSCR survivability scored under realistic lender math.

Operational fragility

High

Sources · Contracts · Emails · Org charts

Patterns in correspondence and structure reveal owner dependence, key-person exposure, and undocumented processes that could stall the transition.

Revenue durability

Med

Sources · Customer lists · Churn cohorts · AR

Invoice data parsed to find concentration, short-term uplift, or route-density risk disguised as "recurring" revenue.

Lender-readiness

Med

Sources · CIM · Lease · Licenses

Cross-references SBA SOP 50 10 8 thresholds against the actual deal structure — flags transferability gaps and SBA-eligibility risks before underwriting.

§ 03 · Use cases

Built for serious buyers, not tire-kickers.

Solo searchers

Kill weak deals in hours. Focus limited time on targets that clear lender math and transferability tests.

Micro-PE & holdcos

Standardize early-stage underwriting across multiple targets. Reduce partner time on obvious misses.

Strategic operators

Pressure-test tuck-ins for integration risk, culture friction, and customer overlap before commitment.

§ 04 · Reference

Answer the objections before you run the scan.

Does AI replace a QoE?: No. It kills bad deals faster and focuses your paid QoE on the targets that deserve it.
What data do I need to start?: A recent P&L, tax returns if available, customer list or AR export, and any contracts you can gather. The more signal, the sharper the scan.
Is my data secure?: Data is encrypted in transit and at rest. We do not reuse your data to train public models.
How fast is the scan?: Most scans complete in minutes. Complex multi-entity uploads may take longer — but still hours, not weeks.
How is this different from generic ChatGPT?: Acquidex AI is methodology-disclosed and cites every signal back to a document section. Generic LLM analysis hallucinates citations; if our model cannot cite, it does not claim.