Building DollarLoop: money decisions a child can own

01 / THE PRODUCT PROBLEM

Learning about money should include deciding what to do with it.

I built DollarLoop around a simple idea: give children room to make money decisions and see the consequences, without opening a brokerage account. Let them research a company, buy a small simulated holding, and watch its value change with actual market prices. Connect that experience to something tangible: work they do at home and rewards their parents agree to.

That sounds like a stock simulator with a chore list attached. The difficult part is the connection. A child needs to understand where money came from, what can be spent, what is invested, and why the total changed. A parent needs to know what they have promised and what still needs their attention.

Even familiar finance conventions can get in the way. A flat chart needs an explanation. A purchase needs to show the cash left over. “Portfolio” needs an introduction, not a replacement with baby talk. A balance can be mathematically correct and still be a bad interface.

The parent decides whether the product belongs in the family. The child decides whether it gets used again.

02 / PRODUCT DESIGN

One family, two different jobs.

For a parent, the useful starting point is practical: assign a chore, approve completed work, see a payout request, and know whether the other parent has handled it. Portfolio performance is interesting, but it should not bury the work that brought the parent there.

For the child, ownership matters. In the investing experience, children can research, buy, sell, or leave cash alone without asking a parent to approve every simulated trade. The adult sets the family’s reward rules; the child lives with the investment choices. Real payouts still require adult approval.

The interface makes earning and investing visible without treating them as the same money. It introduces real vocabulary with explanations: the investments you own together are your portfolio. Research can end with a saved idea instead of a purchase. Short lessons support the decision at hand, rather than becoming a course to finish before being allowed to try.

I want a child to ask for a chore, explain what an ETF holds, or describe why they changed their mind. More trades and longer sessions are not good measures of that.

03 / THE ACCOUNTING MODEL

Practice money still needs real accounting.

DollarLoop keeps the simulated investing account separate from the family’s real-reward ledger. The first tracks cash, holdings, purchases, sales, and market movement. The second tracks earned rewards, requests, approvals, and payments a parent records after settling outside the app.

Consider a hypothetical $1,000 starting balance. Buying $25 of a stock leaves $975 in cash and a $25 holding. Total value is still $1,000. If that holding later rises by 10%, it is worth $27.50 and the account totals $1,002.50. A new chore contribution must not appear as investment performance. A sale must not make the proceeds and the holding count twice.

Those relationships drive both the storage model and the screen. Money uses integer cents; fractional holdings use fixed-precision units. Related changes commit together. Repeated requests must not repeat a purchase or a credit, and two parents approving the same chore must not pay it twice.

The regression checks exercise these less visible cases: retries, competing approvals, transaction failures, and access across family boundaries. They sit alongside browser checks of what a parent or child actually sees after an action.

04 / ARCHITECTURE

A shared market feed, separate family state.

The application uses React and TypeScript, with Cloudflare Workers for server-side logic and D1 for persistent records. The same responsive web app serves a Chromebook, a Mac browser, or a parent’s phone. Permissions belong on the server, not in whichever screen happens to be open.

Market data enters on a separate scheduled path. Daily observations are retrieved and cached centrally, so opening another child’s account does not require another provider request for the same price. Family transactions are then valued against those dated observations. Provider credentials stay server-side.

This fits the lesson better than a streaming terminal. Children can see how their choices changed in value without watching every tick. Price dates and feed limitations remain visible. Caching reduces repeated retrieval; it does not remove infrastructure costs or grant permission to redistribute a provider’s data.

Charts also need an honest starting point. An account with no history should explain that state, rather than manufacture a performance story. Money added to the account and gains from the market need different explanations even when both make the total rise.

05 / DATA AND SECURITY

A child’s session should have a child’s authority.

Parents sign in with Google or Apple. Children use a scoped PIN experience, with family/device setup handled by a parent. A short PIN is a convenience within that relationship, not a substitute for the parent’s identity. Server-side authorization checks the actor and family before allowing access or changes.

Co-parents retain separate identities, so an approval has an accountable actor. Child sessions cannot become parent sessions by changing a tab. Family membership, session revocation, and access checks are part of the application model, not just navigation rules.

The privacy promise must also match the product. DollarLoop stores information needed to run the family experience: profiles, chores, learning progress, and simulated investments. Claiming “we store no data” would be false. The useful work is minimizing what is collected, restricting access, supporting deletion, and preventing unnecessary data from reaching external services.

The public product explains its boundaries: practice dollars, no real trades, no custody of family funds. Those are architectural choices as well as disclosures.

06 / AI

Give the model a small, useful job.

The core experience does not need an AI subscription. Authored lessons and built-in teaching guides work on their own. Optional AI helps a parent prepare an explanation or a practical activity to discuss with a child.

A parent can connect an OpenAI or Anthropic API key. The teaching-card request sends authored educational material and an age band, not a child’s name, portfolio, family records, or open-ended personal conversation. Stored keys are encrypted and bound to the parent account; generation runs on the server with request limits and bounded responses.

The output must fit a small structured card. It has no tool access to trade, approve a chore, move money, or change a balance. A provider failure leaves the built-in guide available. This is an optional aid for an adult, not an AI companion for a child.

That division lets AI contribute where wording and examples help, while prices, calculations, permissions, and state changes remain deterministic. A fluent explanation should never become the source of financial truth.

07 / RELEASE NOTES

What is live, and what I am still testing.

DollarLoop is a live web beta with parent sign-in, child access, chore and reward workflows, simulated stock and ETF investing, and age-matched learning. The source is private. This is a browser release, not an App Store launch.

The current market feed provides dated daily observations rather than real-time quotes or an official closing-auction price. It can include extended-hours activity; split adjustments and dividend credits are limitations of the present simulation. Those details matter before calling any chart a complete investment return.

The next important test is another unassisted session with a child. Can they explain where their cash went? Can they find the job they finished? Do they understand that a price can fall, and that waiting is a decision too? Passing automated checks gets the software ready for that test. It does not answer those questions.

Try DollarLoop, or read the other build notes for more on the product and engineering decisions behind these projects.

QUICK ANSWERS

DollarLoop, in brief.