Issue 2 · 2026-W44 · 26 October-1 November 2026
The best of the first 111: five rows that survived the test
No new rows cleared the test this week, so this issue is a best of: the five rows in the first 111 that carry the clearest figure and the most transferable shape. Every number is the one its source published, and every source is linked.
-
Run a faceless AI YouTube channel to $200k/yr on volume, not on AdSense alone
A faceless video channel run by a tiny team, where the year's revenue came from volume and attention rather than one sponsor.
- Reported
- $200,000 across 12 months
- Split
- ads and sponsorships, per the breakdown
- Demand. A year of the channel's own numbers: $200,000 across twelve months, split between the ad programme and sponsors. It does not say which line did the work.
- Why it could work. Attention is the payer, so one video that lands lifts the month instead of one invoice. The edge is a topic list and a cadence.
- What makes it buildable. Voice, footage and editing are rented per project, which is how a small channel keeps a faceless cadence without a studio.
- Who is already there. Thousands of faceless channels already work every profitable topic, and the ones ahead hold the subscribers and the sponsors.
- What kills it. Platform policy is the single point of failure: demonetisation or a change to the ad programme can halve a year with no notice.
- Cost to test. A model subscription and an editing tool, plus a fortnight of daily uploads on one topic. It counts if a video reaches people unaided.
-
Charge small businesses $1.5k-$3k to wire up AI lead follow-up on Zapier and GPT
A project-based service that wires AI lead follow-up into small businesses on no-code tools, quoted per build rather than per month.
- Reported
- $75K in client work, per the poster
- Price
- $1.5k-$3k per build
- Demand. Owners paid for installed work, not for a course. What the post cannot show is recurrence, because the model is project-based by design.
- Why it could work. A fixed-scope build is easier to sell than a retainer: the owner sees what lands and when. The next job comes from the reference.
- What makes it buildable. A no-code automation platform plus a model API connects phone, inbox and CRM without custom code, so one operator can deliver.
- Who is already there. The same pitch reaches the same local trades from other consultants, and the platforms sell it through their own partner networks.
- What kills it. Scope creep turns a fixed build into unpaid work, and project revenue does not compound: the calendar has to be refilled every month.
- Cost to test. A no-code subscription and two weeks of local outreach. It counts if an owner pays a deposit before any work starts.
-
Valentin tears down SEO Bot, an SEO SaaS he says runs at $93k MRR, to show which pricing-page step actually converts
An SEO subscription that the teardown puts at $93k MRR, sold through a pricing page that shows only the cheapest tier.
- Reported
- $93k MRR, per the author's teardown
- Pricing
- one visible tier, then an upgrade once the data matters
- Demand. A public claim, not an audit. What it does show is that one narrow SEO job, priced visibly, can carry a subscription of this size.
- Why it could work. The cheap tier is the door and the data is the lock: the buyer upgrades when the report says something they cannot get free.
- What makes it buildable. A crawler of public search data, a model to summarise it, a dashboard, and a billing provider for the tiers. The build is integration.
- Who is already there. SEO suites own the category comfortably, and the real substitute is the search console already open in another tab. The teardown names no rival.
- What kills it. The product leans on somebody else's data access: a change in access or ranking rules removes its input. The cheap tier also caps revenue.
- Cost to test. Two weeks and no cash beyond a payment account: build the cheapest tier for one query set, then watch whether a user upgrades.
-
Sell a $297 lifetime membership to a school that teaches non-coders to build AI agents
A paid learning community that teaches non-coders to build AI agents, funded by one-off lifetime memberships instead of renewals.
- Reported
- $225k in 6 months
- Price
- $297 lifetime membership
- Demand. Cash up front and no repeat: the money over six months came from lifetime seats, so a new cohort has to be found every month.
- Why it could work. A lifetime price removes the renewal decision and collects the whole relationship at once, which suits buyers who distrust subscriptions.
- What makes it buildable. A community platform, a course host and a payment processor that takes one charge: a two-person team can open without engineering.
- Who is already there. The promise is everywhere and a few programmes have audiences nobody can match. The real competitor is free: the same lesson on video.
- What kills it. The cohort treadmill: lifetime buyers never renew, so the room has to be filled again every month or revenue stops. Discounting follows.
- Cost to test. A community platform and a payment account. Teach in public for a month before selling seats; it counts if a cohort pays in full.
-
Sell structured workflow and AI-agent software to operations teams through referrals only
Workflow and agent software sold to operations teams through referrals and direct relationships, priced per seat, never launched publicly.
- Reported
- $3M a year, per the founder
- Motion
- referrals and direct sales, no public launch
- Demand. The evidence is a claim: no customer count, price or churn is published. It shows operations software can be sold without a launch.
- Why it could work. Operations buyers are conservative and move on a peer's word, so one reference is worth more than a campaign. The edge is the relationship.
- What makes it buildable. Model APIs and a hosting stack let a small team build workflow tooling; the customer's own systems supply the data, so it is integration.
- Who is already there. Workflow vendors already sell to operations teams, and other agent platforms run the same direct motion. The substitute is a spreadsheet.
- What kills it. Concentration: a referral pipeline produces few large accounts, so losing one changes the year. The sale also takes months, which starves cash.
- Cost to test. Hosting and a model subscription. A quarter of referral conversations and one pilot; it counts if an operations team buys on a reference.
The six-part brief is the paid row
Each pick above is the short version: what it is, the facts the source states, and a line each on demand, why it could work, what makes it buildable, who is already there, and what kills it. The paid index adds the cost to test, the competition in detail, and the risks priced.
Get the next issue by email
Free, weekly, five plays, one click to unsubscribe.
The address is used for this letter only.