Aimone Advisory Book a call
← All work
Online store · my own

TinkerTown: a one-person store that keeps itself current

My own collectibles store. Scheduled AI tasks handle the upkeep that usually sinks a small store, and I approve anything public.

The TinkerTown online store

What changed

The situation

Small online stores rarely fail for lack of demand. They fail from neglect: listings go stale, search rankings slide, nobody sends the email, and the books are a mystery until tax time.

What I built

A standard store, on purpose

Checkout, payments and inventory run on Shopify. Custom code here would only add risk.

A written operating manual

How the store is run: catalog rules, listing standards, compliance limits, marketing voice and the books.

Daily, weekly and monthly routines

They follow the manual: catalog cleanup every day, Google listings and marketing drafts every week, and the books every month.

Gates

Anything that publishes, makes a product claim or spends money waits for my approval.

Why this matters for your business

Most of the value is in the routine, not the software. If you can write down how a job should be done, a scheduled routine can do most of it and hand you only the decisions.

How it fits together

Not everything should run on AWS. Here the right design is a mainstream store platform plus a disciplined operating layer on top of it.

TinkerTown operating architectureOUTSIDESTORE PLATFORM · SHOPIFYAI ROUTINES · GATEDCONTROL1234567Distributorsdealer accounts,stock + pricingShopperssearch, social,directStorefrontcheckout, orders,inventoryCatalog + listingsproducts, feeds, SEOMarketingemail + socialLedgercosts vs salesDailycatalog hygieneWeeklySEO backlog,marketing draftsMonthlyledgerOperating manualthe written rulesCompliance gateslicensing, claimsOwnerapproves changes

Colors group the parts by kind: store, AI routines, controls. Grey is outside the store. Swipe the diagram sideways.

Step by step
  1. Stock and pricing come in from distributor dealer accounts.
  2. Shoppers arrive from search, social and direct visits, and buy through the standard storefront.
  3. The daily routine cleans up the catalog: titles, tags, images and product rules.
  4. The weekly routine works through the search-listing backlog and drafts marketing.
  5. The monthly routine reconciles the ledger, comparing costs with sales.
  6. Every routine reads the operating manual first. The manual is the single source of how the store runs.
  7. Compliance gates hold anything risky, and I approve what ships.

Technical details

For your IT person: components, trade-offs and security. Open any section.

Components
LayerServiceWhat it does here
CommerceShopifyStorefront, checkout, payments, inventory and product feeds.
RulesOperating manual (written, versioned)Catalog standards, listing rules, compliance limits, marketing voice and ledger steps. The routines treat it as their instructions.
OperationsScheduled AI routines (Claude)Daily, weekly and monthly sessions that read the manual, act through the store connector, and log what they did.
ControlsApproval gatesAnything that publishes, makes a claim or spends money needs owner approval. Everything else is reported, not asked.
BooksLedgerA monthly reconciliation of sales against costs and subscriptions.
Design decisions and trade-offs
Buy the commodity, build the edgeCheckout is solved. The edge is consistent upkeep, so that is where the build effort went.
Manual first, prompts secondThe rules live in plain language a person can read and edit. Changing how the store runs means editing the manual, not code.
Cadence matches the workCatalog drift is daily, search work is weekly, and books are monthly. Each routine is small enough to check.
Report by default, ask by exceptionRoutine fixes happen and get logged. Only risky or irreversible actions stop for a yes.

Want your store or practice to run like this?

If you can describe how the work should be done, I can usually make most of it happen on a schedule. Book a free 30-minute fit call.

More write-ups