Most expense apps either feel like corporate banking tools or push everything into the cloud. For freelancers and young professionals, especially people spending across currencies, tracking receipts is still manual, messy, and easy to abandon. People wanted something that could capture a receipt quickly, work offline and stay private on the device, export cleanly into Google Drive and Sheets, and handle multi-currency without forcing one home worldview on every transaction. The gap was accountability without friction: scan, verify, organise, and actually keep up with spending.

Stanca is a local-first expense companion. Users sign in with Google, capture or upload receipts, review OCR-extracted fields on-device, and save expenses to a SQLite database on their phone. Verified data syncs to their Google Drive (receipt images) and Google Sheets (expense rows), not a Stanca-hosted backend. There is no custom server in v1. The phone is the operational source of truth; Google is the user-owned backup and reporting layer. That keeps the product fast, private, and cheap to run while still feeling connected to a familiar workflow.

• Receipt capture & on-device OCR

  Camera, gallery, or payment-screenshot import; ML Kit parsing for amount, date, merchant, currency; human review before save.

• Expense management

  Manual entry, edit, categorisation, duplicate detection, soft delete.

• Multi-currency

  Original currency preserved; Home Currency for dashboards, budgets, and summaries; mid-market FX via Frankfurter.

• Budgets & alerts

  Weekly, monthly, and yearly budgets with local notification alerts and in-app inbox.

• Insights

  Dashboard with category breakdown (donut chart), recent activity, sync status.

• Google workspace sync

  Automatic Drive receipt upload and Sheets export with retry queue and offline resilience.

• Recurring expenses

  Pattern-based recurring item tracking.

• Currency converter

  In-app conversion using cached FX rates.

• Privacy centre

  In-app policy, terms, data export (JSON), account deletion, permissions management.

• Premium UX polish

  Custom design system, orbital onboarding, animated splash, success toasts after completed actions.

• OCR you can trust

  Built a multi-stage pipeline: image normalisation, ML Kit, layout-aware parsing, field-specific parsers, confidence scoring, and a mandatory review screen. Nothing auto-syncs before the user confirms.

• Offline-first sync without a backend

  Expenses persist to Drift immediately; a durable sync queue handles Drive upload and Sheets writes with deduplication, exponential backoff, credential refresh, and manual retry-all from Settings.

• Multi-currency without losing fidelity

  Each expense keeps its original amount and currency. Home Currency conversion happens at display and budget time using cached Frankfurter rates, with explicit handling for ambiguous symbols like $ and R.

• Privacy by architecture, not policy alone

  Local-first storage, least-privilege Google scopes, no ads or tracking SDKs, in-app deletion and export, and store-ready compliance documentation without inventing a fake legal entity.

• Premium feel on a solo build

  Invested in motion (splash reveal, success toasts), a cohesive design system, and onboarding that explains the product without overwhelming, so Stanca feels intentional, not like a side-project template.

Shipped v1.0.0+1 codebase with full mobile product path: auth, OCR, expenses, budgets, sync, settings, and privacy. Eight or more milestones documented as complete in repo. Store-ready privacy and legal framework (in-app plus internal compliance docs). GitHub repo with architecture, schema, sync, OCR, and release documentation. In progress: App Store and Play Store submission, hosted legal pages on stancaapp.com, production launcher icons and final store screenshots.

1. 1.Local-first is a product decision, not just a tech choice. Users feel ownership when data lives on their device and in their Google account, not on your server.
2. 2.OCR is only as good as the review step. Parsing gets you 80%; UX around ambiguity, currency confirmation, and duplicate detection gets you the rest.
3. 3.Sync queues need to be boring and reliable. Deduplication, stuck-job recovery, and human-readable errors matter more than clever real-time tricks when there is no backend.
4. 4.Compliance belongs in the build, early. Privacy centre, deletion, export, and store disclosure docs are easier when the architecture already matches what you claim.

Live project

www.stancaapp.com/