A products business — the Adafruit / Pololu path¶

Status: proposal — awaiting decision, 2026-09-15. Angel: "becoming some sort of seller like Framework: value-forms (the twelve forms of value; which ones this proposal covers). Adafruit or Pololu." This is how that works for a two-person shop with its own SMT line, and what the first three products should be. Sits next to first-money (services) and recurring-revenue; task 47 is only the eBay redirect; the broader storefront is a separate end-state plan.

What makes Adafruit and Pololu work (and what we already have)¶

Ingredient	Them	Us today
Own designs, small and opinionated	breakout boards, drivers, carriers	16 vetted KiCad blocks, the tester, the Real Cow boards
Documentation is the marketing	a tutorial per product, the product page teaches	the engineering notes + the note-to-social pipeline; a note per product is the same machine
In-house assembly of small runs	their own lines	Neoden YY1, reflow, the Universal Tester, ERP builds (SOP-006), labels
Sell where the buyers already are	own store + distributors	nothing yet: Tindie, then own store, then Amazon/DigiKey Marketplace
Margin that survives	3–4× parts cost on boards, 2× on kits	to be designed in from the first BOM

The gap is not capability. It is products chosen for a buyer, priced with margin, with a page that teaches — and a channel.

Rules that keep it from bleeding money¶

Sell designs that already exist, first. A product that is one of our blocks on a 25×25 mm board with a header is a week of work; a new design is a quarter.
Price at 3× the landed BOM+assembly cost, minimum, or do not make it. Pololu's price for a small regulator board is $6–15 on a $2–4 cost; that is the shape.
Runs of 50, never 500, until a product has sold 50. Stock is cash on a shelf.
No radios in the first three. A board with a transmitter needs FCC evidence (modular certification or our own test); a board without one needs a CE/FCC statement and RoHS, not a lab. Keep the first products in the no-lab class.
Every product ships with a note (the tutorial), and the note ends with the buy link. That is the whole Adafruit trick, and the content pipeline already does the hard half.

The first three, from what is on the shelf¶

#	Product	From	Why it sells	Cost / price (guess)
1	USB-C PD sink board — 5–20 V negotiated out to a screw terminal, status LED	`blocks/usb_c` + the USB-C PD note	every maker converting a 12/19 V device to USB-C; the note already ranks	$4 / $15
2	1-cell Li-ion charger + soft power switch — charge, protect, one-button on/off, 3.3 V out	`Battery_Charger_1S` + `Soft_Switch` + `Regulator-5_3.3`	the most reused three blocks in the library; a portable-project staple	$3.5 / $14
3	STM32G0 32-pin dev board with our pinout — USB-C, SWD, the blocks' rails	`Stm320b1k_32pin` + `usb_c`	it is what we build on; a board with a real tutorial and a KiCad template behind it	$5 / $18

Number 4, when a radio is worth its paperwork: the NRF24 module block. Number 5, later and bigger: the Universal Tester as a product for other small shops.

Channels, in order¶

Tindie first (free to list, ~5% fee, the exact audience, no store to build). It is also the cheapest way to learn what sells before spending a week on a store.
Own store (separate from task 47) once three products exist and one has sold 50: Stripe checkout on the site, stock from the ERP (the channel adapter, task 172, pushes stock out and pulls orders in — that was the decision on task 81).
Amazon / DigiKey Marketplace only for a product that is already selling; both eat margin and demand stock depth.

What it costs to start¶

Product 1 from block to first run: ~1 week design + review (SOP-021), $150 for 5 prototype boards and stencil, $250 for a 50-piece run's parts, ~2 days of assembly and test.
Compliance paperwork for a non-radio board: a declaration and RoHS statements from the part data, no lab. Budget one day.
Tindie: $0. Own store: separate future scope; task 47 only redirects to eBay.
Cash tied up at any time with three products in runs of 50: roughly $1,500.

The order of work¶

Angel picks the first product (recommendation: the USB-C PD sink — the note is live and ranking, so the tutorial already exists).
Design from the block in a product repo from Project_Template (SOP-021/022), 5 prototypes.
Product record in the ERP (class A part, BOM, build order), the tester plan, the label.
The product note (tutorial) + the Tindie listing, both from the same text.
First run of 50; sell; watch. Then product 2.

What Angel decides¶

Which product first, and whether the price shape above is acceptable.
Tindie as the first channel (recommended) or wait for the own store.
Whether product design time comes out of client hours or evenings — that is the real cost.

Log¶

2026-09-15: written.
2026-09-17: Corrected task 47 references against the live register: eBay redirect only, not a full storefront.