Product development · privacy engineering · iOS

Investment context, designed with privacy at the boundary.

InvestCircles explored how portfolio activity, market evidence, and trusted discussion could live in one product without making financial information public by default.

The active product has been retired. This site is a static professional case study: no accounts, live data, broker connections, forms, analytics, or background services.

The product thesis

“Context should make a portfolio easier to understand—never easier to expose.”
01
Evidence before explanations
02
Visibility enforced at the data boundary
03
Background work kept bounded and auditable

Static portfolioNo application backend

Synthetic mediaNo real accounts or balances

Honest scopeNo adoption or return claims

01 · Product

A daily brief, not another balance screen.

The product connected holdings with reported earnings, market events, transactions, and completed trades. When a catalyst could not be verified, the interface was designed to say so.

Synthetic InvestCircles product screens showing a portfolio brief, an earnings comparison, and an event-linked price chart.
Synthetic product media. Figures and company names are fictional.

02 · Privacy

Social investing with explicit visibility rules.

Owner, friend, circle, and stranger access contracts shaped the data model. Holdings, percentages, amounts, identity, and portfolio activity could be shared independently instead of hidden only in the interface.

Synthetic InvestCircles screens showing private circles, a fictional group conversation, and selective portfolio visibility controls.
Synthetic people, messages, balances, and connected accounts.

03 · Architecture

One product, separated approval gates.

Web, iOS, data, workers, and distribution were treated as distinct boundaries so each could be tested, deployed, or retired independently.

Clients

Web + Capacitor iOS

Untrusted interfaces with a shared TypeScript product surface.

Application

Next.js 16 + React 19

Interactive product routes, server APIs, and progressive bootstrap.

Data boundary

Supabase Postgres

RLS, explicit RPC contracts, Auth, Storage, and Realtime.

Background

Signed Edge workers

Timestamped requests, nonces, leases, cursors, and idempotent writes.

Privacy

Protect reads at the boundary

Financial visibility was a database contract, not a CSS decision.

Evidence

Canonicalize before notifying

Contradictory earnings actuals were marked disputed and blocked from alerts.

Performance

Bootstrap the shell once

Session, theme, minimal profile, and portfolio summary shared one initial response.

Cost

Bound retention and work

Detailed market snapshots rolled into compact daily aggregates.

04 · Dated evidence

Local checks before public claims.

These results describe the 10 August 2026 local verification run. They demonstrate engineering discipline—not users, revenue, investment performance, or production reliability.

194/194TypeScript testsPassing locally
85Static pagesGenerated by the local production build
8/8Public-flow checksChromium + iPhone WebKit
Debug + ReleaseiOS Simulator buildsUnsigned with Xcode 26.6
0 knownDependency vulnerabilitiesModerate severity or above at the dated audit; not a security guarantee
InvestCircles engineering evidence card listing dated local test, build, browser-flow, iOS compilation, and dependency-audit results.
Evidence scope and AI-assistance disclosure are preserved in the artifact.

Current status

Preserved as engineering work, retired as a live product.

App Store distribution and TestFlight access were withdrawn. The production backend and scheduled work were shut down through a reversible closure process, while source, migrations, backups, domain, and platform identifiers were preserved.

The iOS application was built and tested in Simulator but was not released in the App Store. All interface media on this site is synthetic and contains no real user, broker, balance, trade, message, token, or account data.

Authorship disclosure

Directed, reviewed, and evaluated by Daniel Peña Hartwig.

The concept, product rules, privacy boundaries, workflows, and release criteria were directed by Daniel Peña Hartwig, with substantial AI-assisted implementation, review, testing, and documentation. This case study does not imply that every line was authored manually.