Which business should a solo operator launch with under $5,000, 30 days to first revenue, strong in-house design and paid-ads capability, and no licensed credentials? Every candidate shares one shape: upload documents, get a valuable report back, pay once with a guarantee, subscription optional.

How candidates were judged

Each idea scores 1–5 on ten weighted criteria — profitability, upfront cost, order value, demand growth, underservice, social proof, ad reachability, word-of-mouth, subscription fit and speed of cheap invalidation — then takes a penalty for regulatory risk. Before scoring, every idea faces five kill-filters. The most lethal by far: is this deliverable already given away free as a sales lead-magnet? That one filter killed 26 ideas, four of them previously top-ten. You cannot charge for what competitors use as bait.

This runs in cycles, and that is the point. An automated loop rebuilds the database weekly from regulatory calendars and enforcement actions, re-screening what it already holds. Two cycles in, the honest result is not a higher ceiling — cycle one still holds the best idea. What cycling buys is correction: cycle two found a funded competitor in a gap cycle one called empty, retracted a regulatory deadline that never arrived, and killed 9 of the previous top 20. Desk research decays; a static report would still be repeating both errors.

The three to build

Runners-up

The Next 7

Ideas that survived every kill-filter but lost the top three on score

These are not rejects — every one cleared all five kill-filters. They sit just below the top three, and the gap is often smaller than the margin of error in the scoring. One is the natural second product for pick #1. The rest of the active set sits below these in the Leaderboard.

Score = weighted rubric result out of 100 after the risk penalty. Risk 1 = no licensure exposure, 5 = effectively requires a licence. Tier = evidence level: T0 unverified prior T1 screened against the kill-filters T2 competitor pricing and demand statistic sourced T3 adds live ad-cost data — nothing has reached T3 yet. Excluded here: every candidate killed on evidence. Nothing is unscreened. Full database on the site.

Construction & trades — the same buyer as pick #1

Bid Leveling & Comparison Report87.6

GC or building owner · $199–$899 · $299/mo · risk 1/5 T2

Owners comparing 4–8 subcontractor bids have no cheap way to normalise scope. Same upload motion as pick #1 — this is SKU 2, not a second company.

Regulated & compliance-driven — mandated documents

Supplement & Health Claim Substantiation Dossier82.8

Supplement / wellness brand · $1500–$7500 · $399/mo · risk 3/5 T2

Published price band of $1,500–$7,500 per claim set, FTC penalties above $53,000 per violation, and ad lawyers dislike document assembly.

TTB COLA Label Pre-Submission Rejection Risk Review79.5

Craft brewery / distillery / winery · $300–$900 · $149/mo · risk 2/5 T2

Small producers submit dozens of labels a year and every change needs a fresh COLA, while beverage lawyers bill $350–$600/hr. Rejections delay launches by weeks. Nobody serves that volume at a per-label price.

Provider Enrollment Revalidation & Credentialing Application Audit79.2

Small practice / new provider · $400–$1500 · $149/mo · risk 1/5 T2

Missing a Medicare revalidation deadline deactivates billing privileges and stops payment outright. Platforms are priced for volume and health systems do it free for employed providers — the independent practice with an episodic need is the gap.

ATF FFL Bound Book & 4473 Discrepancy Audit78.3

Gun shop / FFL holder · $900–$3000 · $249/mo · risk 2/5 T2

ATF revokes licences for wilful recordkeeping violations, so the bound book must be inspection-ready at all times. NSSF subsidises on-site visits for Premium members, but most small FFLs are not members, and a remote document audit is a different product from a site visit.

Money the buyer is already owed

CAM / Operating Expense Reconciliation Audit83.1

Commercial tenant · $499–$2499 · $299/mo · risk 2/5 T2

~40% of reconciliations contain material errors and recovery averages 15–20% of CAM billed. Incumbents take 30–50% contingency; a flat fee where the tenant keeps everything is the wedge.

Other shapes worth a look — different buyers, different rhythms

Tariff Exposure & Mitigation Report79.6

Importer / manufacturer · $999–$4999 · $499/mo · risk 3/5 T2

Tariffs are a dominant 2026 operating issue for importers and manufacturers, and contractors are already re-bidding around the volatility. Urgent, expensive and poorly served at the SMB importer level — but the riskiest card here at 3/5.

What each would have to do to matter Not a forecast — a break-even model

Per order	Construction Takeoff	HOA Review	Grant Readiness
Blended price	$499	$189	$399
AI + payment fees	−$23	−$11	−$22
Gross per order	$476	$178	$377
Orders per month needed to net $10,000
If CAC is low	29	80	35
If CAC is mid	38	108	42
If CAC is high	135	never	94
Capacity and shape
Minutes to fulfil one	30	25	25
Solo ceiling / month	240	288	288
Net at mid CAC, if volume is reached	$123k/yr	$120k/yr	$120k/yr

Read the middle block, not the bottom one. All three reach roughly the same income if they reach volume — so the real question is which volume is believable. 38 takeoffs a month from small subcontractors is a plausible year-one business. 108 condo reviews or 42 grant reviews a month is a different job. None of the three has the high-ticket cushion the killed WISP offered: all three go permanently underwater if acquisition proves expensive, which is why the high-ticket bet now sits with CAM Reconciliation and Supplement Substantiation in the test order (see the Proforma). CAC is the single unknown that decides all three, and no test has been run yet. One point favours all three: payment is taken at or before delivery, so there are no receivables and no working capital tied up — which is why $3,000 of testing is enough to learn something.

Two things this table cannot score. Whether frontier AI commoditises the deliverable — defensible, but only through choices made on day one, which the playbook covers. And what the work feels like: 38 contractor conversations a month is a different life from 108 anxious condo buyers.

What this took — and what it would have taken by hand Estimated, assumptions shown

Work performed	Volume	Manual rate	Analyst hours
Specify each candidate — buyer, input documents, output report, price band	335 ideas	10 min ea.	56
Score every candidate against 10 weighted criteria plus a risk rating	3,350 judgments	8 min/idea	45
Research threads run — search, read sources, extract and record findings	111 threads	25 min ea.	46
Deep verification — named competitors, live pricing, demand statistics	38 candidates	60 min ea.	38
Kill decisions documented with reason and source, plus synthesis, ranking and full write-up	263 kills / 8,800 wds	—	104
Total human effort to reproduce this database			~288 hrs

7 weeks

By hand

36 days

Full-time equivalent

$43,300

At $150/hr analyst rate

~6 hrs

Actual elapsed

48×

Compression

Assumptions: manual rates assume an experienced analyst working from a brief without prior domain knowledge across 18 industries. Volumes are counted from the database and research log, not estimated. $150/hr is a mid-market consulting benchmark. The database refreshes weekly at no additional analyst cost — a recurring saving this one-off comparison does not capture.

The ten questions asked of all 335 candidates	Weight	Construction Takeoff	HOA Review	Grant Scoring
Will it make money quickly?	15%
Can it start cheaply?	10%
Is each sale worth enough?	13%
Is demand big and growing?	12%
Is the need underserved?	12%
Do buyers complain publicly?	7%
Can we reach buyers with ads?	11%
Will customers refer others?	9%
Does a subscription fit?	5%
Can we test it cheaply?	6%
Legal or licensing risk (lower is better)	penalty	1 of 5	2 of 5	1 of 5
Final score out of 100		91.8	84.5	83.6

AI Document–Report Businesses

The question

How candidates were judged

The three to build

Confidence — read honestly

The Next 7