The first time a buyer sees an AI-generated deal memo, the reaction is usually some version of: "This is actually pretty good." The second reaction, a few weeks later when a deal they thought was solid falls apart in QoE, is: "I trusted it too much."
Both reactions are correct. AI deal memos are genuinely useful—and genuinely limited. The buyers who use them well understand exactly where the line is.
What AI Gets Right
Financial Extraction
Pulling numbers from a CIM is tedious and error-prone when done manually. AI does it well: revenue, SDE, add-backs, gross margin, YoY trends. A good AI memo will calculate the implied multiple automatically and flag if it's outside normal ranges for the sector.
This alone saves 30–45 minutes per deal. In a high-volume search, that compounds quickly.
Structural Red Flags
AI is excellent at spotting the patterns that appear repeatedly in troubled deals:
- Revenue concentration (one customer >20% of total)
- Declining revenue trends masked by add-backs
- Lease expiration within 2 years of proposed close
- Unusual seller financing requests that may signal bank ineligibility
- Equipment that's fully depreciated but operationally critical
These aren't things you'd miss with careful reading—but careful reading takes time you often don't have when evaluating 20 deals a week.
Debt Service Analysis
Most buyers have a rough sense of DSCR math. AI formalizes it: given SDE, purchase price, down payment %, and a 10-year SBA 7(a) loan at current rates, does this deal pencil? An AI memo that shows DSCR of 1.12 at ask is telling you something important before you've spent a minute on due diligence.
What AI Gets Wrong
Owner Dynamics
The most important variable in any small business acquisition isn't on the income statement. It's the relationship between the owner and the business. Is the owner leaving because they're burnt out, or because the business is burning? Is the "transition period" offer genuine, or is the owner desperate to extract and disappear?
AI can't read this from a CIM. A CIM is a marketing document—it's written to sell the business, not to reveal its vulnerabilities. Qualitative owner dynamics only emerge from conversations, reference checks, and sometimes just gut instinct from sitting across a table.
Operational Reality vs. CIM Reality
CIMs describe businesses as they should work, not as they do work. The "documented processes" section of most CIMs describes processes that exist as Google Docs but are never actually followed. AI will summarize what the CIM claims; it won't tell you whether those claims are true.
A QoE (Quality of Earnings) report from a transaction advisory firm is the formal answer to this. But QoEs cost $15–30K and are done after LOI. The informal answer is customer calls, employee interviews, and vendor reference checks—none of which AI can do for you.
Industry Context
AI can apply generic acquisition heuristics. It doesn't know that HVAC businesses in the Sunbelt trade at premium multiples right now, or that certain staffing niches are getting compressed. Sector-specific knowledge comes from brokers, industry conferences, and talking to other buyers—not from a CIM.
The Right Workflow
The buyers who use AI deal memos effectively treat them as a first filter, not a final answer. The workflow looks like this:
- AI memo generation: Upload CIM or paste listing, get structured memo in under a minute. If the AI memo doesn't pass your criteria, move on—no human time spent.
- Human review of flagged items: Read the memo's risk section carefully. For any flag you disagree with or want to investigate, note it explicitly before the broker call.
- Broker call armed with memo: You're not starting from scratch—you're probing specific items. "The CIM shows a 22% customer concentration with one account—can you tell me more about that relationship?" This is a different quality of conversation than a generic introductory call.
- Go/no-go decision: Most deals die here, after the broker call. The memo surfaced the questions; the call answered them. If the answers are unsatisfying, the memo gave you the context to articulate why.
- LOI if it survives: Now the real due diligence begins. The AI memo becomes a living document—updated as QoE results and management interviews produce new information.
SBA-Specific Considerations
SBA 7(a) financing adds a layer of scrutiny that AI memos should account for but often underweight. Specific flags for SBA deals:
- Life insurance requirements: SBA lenders typically require key-man life insurance equal to the loan amount. This adds cost and occasionally reveals health issues that affect deal structure.
- Standby seller notes: If seller financing is structured as a standby note (no payments during SBA loan term), it can still count toward SDE, but the documentation requirements are specific.
- Injection requirements: SBA 7(a) typically requires 10% equity injection from the buyer. AI memos should model this explicitly so buyers understand their actual cash requirement at close.
- Goodwill threshold: SBA has specific rules around businesses with >50% intangible/goodwill value. Most service businesses qualify, but it's worth flagging when goodwill is especially high.
DealPacket's AI memo engine includes SBA-aware analysis by default—DSCR calculated at current 7(a) rates, injection requirement flagged, and standby note implications noted where relevant.