The Fleet

← All digests

Digest — The Entrepreneur, 2026-09-02

Escalation bar crossed on two counts: the overall verdict flipped in kind, and the SBLOC/SBA financing questions got answered. But neither is the thing you most need to read.


1. Two minutes of your time, and it is the highest-value thing in this domain

Charter v3 never made it into the Drive folder. You wrote it on 31 August — the one that converts this agent from category search into a scanner for specific businesses listed for sale. It was handed to you as a file to drop into Fleet — Agent Substrate/charters/. It is not there. The folder holds entrepreneur.md dated 2026-08-07, which is v2.

So the Wednesday trigger read v2 this morning, and will read v2 again next Wednesday. This agent spent the run chartered to eliminate-and-validate while you had asked it to scan deal flow.

Drop the file in. Newest-wins, no trigger edit needed.


2. This domain was wrong about its own headline question for three and a half weeks

Three consecutive run sections of business--candidates.md called your risk answer under the uncapped SBA personal guarantee "the binding constraint," and reported it open and unmoved.

You answered it on 8 August. You said you weren't worried about the guarantee exposure — good credit, a good plan gets the money — and told that session to skip the lender call and assume financing. Runs 4 (19 Aug) and 5 (26 Aug) both opened by reporting it open. Both were wrong when they were written.

The cause is not subtle: this agent read the corpus and did not read the record of what you had actually said. Recorded in full, because it is the second time this domain has been wrong about its own headline question for the same reason.

One flag, once, and then it's yours. "Assume financing" was reasonable against what you knew on 8 August. Since then the corpus has established three things you didn't have: change-of-ownership leaves the streamlined 7(a) Small Loan process entirely on 1 October; the DSCR floor for a first acquisition rises to 1.25× on historical earnings with projections barred; and the corpus's own median scenario computes to 1.05× and fails that floor — and fails the floor in force today. Not reinstating the lender call on my own authority. Just putting those three facts next to the instruction.


3. Your Vector bar turns out to be the thing that kills deals — not the SBA

You set a new bar on 31 August: a venture has to beat Vector, so ~$21,600/year. Worked into arithmetic against this corpus's own financing assumptions (10-year term, 9.75%, 90% LTV → annual debt service of 14.12% of purchase price):

Lender line: SDE − replacement labour ≥ 17.65% × price (the 1.25× DSCR floor) Vector line: SDE − replacement labour ≥ 14.12% × price + $21,600

They cross at a purchase price of about $612,000. Below that, the Vector line is tighter. Your entire modelled band is $250K–$600K.

So across every deal this domain has been contemplating, the constraint that kills it is your own opportunity cost, not the regulator. Three consecutive runs of SBA financing research were sharpening the wrong blade.


4. The scanner direction is technically viable, and your screening rule is a pre-built filter

Tested this run because v3 depends on it and nobody had checked. BizBuySell's Chicago Metro Area absentee-run filter and BusinessesForSale's Chicago search both returned structured listings — price and disclosed cash flow — to a plain fetch. BizBuySell and BizQuest both publish dedicated absentee-owner filter URLs, and BizBuySell publishes an owner-financed filter too, which is half the deal structure this domain has modelled since run 2.

Ran the screen against fifteen live listings. Two fail before any labour is replaced at all — a tutoring franchise at $200K on $45,000 of cash flow, an indoor play centre at $199,000 on $35,129. Two of the largest fail the DSCR floor outright on disclosed figures.

None of these is a candidate. No listing investigated, no broker contacted, no financials seen. And "absentee-run" is a tag the seller applies — this corpus's own finding is that owner-dependence is the single most misrepresented fact in business listings. The screen says which listings are worth one question, not which are good.


5. The financing answer, and it is good news

Three secondary sources published 16–24 August now describe Appendix 15 of SOP 50 10 8.1, the thing run 5 flagged as putting your whole financing argument back into "unread."

All of this is reported, not verified. No source quotes the primary text, two of the three sell services in this space, and the actual .docx is still robots-blocked to agents and one click for you.


Verdicts


Two questions only you can answer

  1. Does the Vector bar mean money, or money per hour? You phrased it as yield "for the time it takes." The screen currently tests only the money half. A business needing 10 hours a week to clear $21,600 passes the arithmetic and fails the bar as you stated it.
  2. Given the median case computes to 1.05× against the new floor — does "assume financing" still stand?

Entries written: business--sop-50-10-8-1-appendix-15.md, business--deal-screen.md. Updated: business--candidates.md (35,738 → 58,944 B). Deliberately not rewritten: business--land-flipping.md and business--sba-financing-mechanics.md — neither was read in full this run, under the rule the 2026-09-01 run established after a truncation defect.