L
LAMBOAPP
Sign in
Concepts·3 min read

AI scoring — how Claude reads every listing

Claude reads every listing, writes a thesis, flags the obvious scams, and assigns a fit score. It is not due diligence.

AI scoring

Every listing that enters the pipeline gets read by Claude before it ever appears in the feed. The output is three things:

  1. A one-paragraph thesis — what this business is, why someone might want it, and what the plausible path to a return looks like.
  2. Red flags — the obvious tells that this is a scam, a distressed pump, or a numbers-don't-add-up situation.
  3. A fit score — how well this matches the archetype of "boring, cash-flowing, sub-$5M SDE, not a tech gamble."

What "red flag" means

The scorer is looking for the mechanical tells that recur across bad listings:

  • Revenue "since inception" that doesn't reconcile with the age of the business.
  • SDE that's a suspiciously high fraction of revenue (>60% is unusual for real service businesses).
  • Broker language that overuses "turnkey," "absentee," "scalable" without any of the numbers those words imply.
  • Missing lease terms on brick-and-mortar deals.
  • Amazon/Etsy dependency where the platform account isn't part of the sale.
  • Founder-in-the-picture businesses being sold as "systems in place."

It's not exhaustive and it's not a substitute for reading the listing yourself. It surfaces the obvious cases so you can spend your reading time on the deals that survived a first pass.

What the fit score is (and isn't)

The fit score is a scalar 0–100 that answers one question: given the archetype of a solid acquisition target, how well does this listing match? Higher is better. The archetype is roughly:

  • 5+ years established.
  • Sub-$5M SDE.
  • SDE multiple in the 2x–4x range.
  • Not obviously dying (declining revenue trend, single-customer dependency, dead industry).
  • Real operational moat (not just a domain name).

It is not a buy signal. It is a "worth 15 minutes of your reading time" signal.

What it doesn't do

  • It doesn't do financial due diligence. No P&L reconstruction, no working-capital normalization, no tax analysis. That's the QoE analyst's job after you're under LOI.
  • It doesn't judge the operator side. Whether you are the right person to run a laundromat is out of scope.
  • It doesn't predict outcomes. Every past acquisition failed to hit projections; every future one will too. The score is a filter, not a forecast.

Failure modes

The scorer will occasionally:

  • Miss a red flag because the listing hides it in a photo caption or a broker-supplied PDF.
  • Flag a legitimate business as risky because the broker copy is unusually promotional.
  • Score a niche business (e.g., a specialty machine shop) as "low fit" simply because the archetype prior is skewed toward service SMBs.

The right way to use the output: read the thesis, read the red flags, then form your own opinion. Don't outsource the reading — outsource the filtering to find things worth reading.