Due Diligence in the Age of AI: What's Changed for LMM PE

Due diligence at a lower middle market PE firm has always been a different animal than at a large-cap fund. The targets are smaller, the data is messier, the management teams are less sophisticated, and the legal budgets are tighter. A $50M deal doesn't get the 20-person diligence team and $2M in advisory fees that a $500M deal commands. LMM DD is done by two to four people, often on a 60–90 day timeline that leaves little margin for delays.

AI is particularly impactful at the LMM level precisely because the resources are constrained. The large-cap firm can throw bodies at a complex DD workstream; the LMM firm can throw AI at it for a fraction of the cost and comparable quality on specific tasks.

Document Review: Where AI Delivers the Clearest ROI

The document review phase of DD — processing NDAs, customer contracts, employment agreements, IP assignments, and regulatory filings — is the most time-consuming and the most amenable to AI acceleration.

Kira Systems remains the gold standard for structured contract extraction. It identifies and extracts key provisions (change of control clauses, non-compete terms, IP assignment language, termination provisions) from large document sets with high accuracy. Primarily accessed through law firm relationships — most major M&A firms have Kira licenses that benefit their PE clients.

Luminance offers a more flexible approach, particularly useful for varied document types that don't fit neatly into pre-defined extraction templates. Its anomaly detection — flagging provisions that differ materially from the norm across a document set — is particularly valuable in DD for identifying non-standard terms in customer or supplier contracts.

Harvey represents the next generation. Built specifically for legal AI, it handles complex multi-document analysis, can cross-reference provisions across agreements, and generates memoranda-quality summaries. For PE DD, Harvey's ability to synthesize across a full data room — not just extract from individual documents — is the differentiator. Primarily available through law firm partners currently.

For an LMM deal with 200–500 documents in the data room, AI document review tools reduce the outside counsel time on document review by an estimated 30–50%. On a $150K legal budget, that's $45K–$75K in savings — often enough to justify the AI tool cost several times over.

Financial Analysis: Faster but Still Human

AI is useful but not yet trustworthy for financial DD. The applications that work:

• Quality of earnings analysis acceleration: AI processes historical financial statements, identifies non-recurring items and adjustments, and generates a preliminary normalized EBITDA bridge. The QofE advisor still does the work — but starts from an AI-generated first draft rather than a blank template. Time savings: 2–3 days.
• Working capital analysis: AI processes monthly AR, AP, and inventory data to identify seasonality patterns, aging trends, and potential working capital peg issues. Particularly useful when the target's books are in QuickBooks and the data export is a CSV that needs significant cleaning.
• Variance analysis: AI compares management projections to historical performance and flags assumptions that appear inconsistent with trailing results. "Management projects 18% revenue growth, but trailing 3-year CAGR is 7% with no identified new growth drivers" — the kind of flag that AI catches systematically.

What AI cannot do in financial DD: evaluate the quality of the management team's financial controls, assess the reliability of the company's accounting practices, or make judgment calls about the sustainability of customer relationships. These require an experienced diligence professional on-site.

Market Research and Competitive Analysis

AI dramatically accelerates the market research component of DD. Tasks that previously required an analyst to spend days synthesizing industry reports, competitor information, and market sizing data can now be processed in hours:

• Market sizing: AI processes industry reports (IBISWorld, Freedonia, public company filings) to generate a bottom-up market sizing analysis for the target's specific niche.
• Competitive landscape: AI compiles competitor profiles from public data — websites, press releases, job postings, SEC filings — and generates a competitive positioning summary.
• Customer and supplier research: AI processes public information about the target's key customers and suppliers to assess concentration risk and relationship stability.

Background Research and OSINT

AI-assisted open source intelligence (OSINT) has become a standard part of management DD. AI tools can process public records, news archives, social media presence, litigation databases, and professional history to build comprehensive background profiles on key management team members. The output supplements — but doesn't replace — professional background checks and in-person management meetings.

The key ethical constraint: AI background research must comply with the same legal and ethical standards as traditional background checks. Processing publicly available information is acceptable; invasive surveillance or processing protected information is not, regardless of technical capability.

LMM-Specific Considerations

LMM deals present unique DD challenges that AI must accommodate:

• No virtual data room: Some LMM sellers don't set up a formal VDR. Documents arrive via email, Dropbox, or even physical handoff. AI tools need to process whatever format the data arrives in.
• Messy books: Cash-basis accounting, personal expenses commingled with business, incomplete records. AI can help normalize this data but requires more human oversight than clean-book scenarios.
• Owner-dependent businesses: The seller's personal relationships with customers, suppliers, and employees are often the business's most critical asset. No AI tool can evaluate this — it requires the GP's judgment from in-person assessment.
• Compressed timelines: LMM deals often move faster than large-cap deals because sellers are eager to close and the buyer universe is competitive. AI's speed advantage is most valuable when the timeline is tightest.

The DD process will always require experienced human judgment. What changes with AI is where that judgment gets applied — to interpretation and decision-making rather than data assembly and document processing. For LMM PE firms where every team member's time is precious, that shift is transformative.