Skip to content
arudaev edited this page Apr 5, 2026 · 6 revisions

Zelara Core - Technical Documentation

Welcome to the technical documentation for Zelara core.

Quick Navigation

Project Status

Phase: Active Implementation ✅

All major decisions made:

  • ✅ Technology stack: React Native + Tauri + TypeScript + Rust + ONNX Runtime
  • ✅ MVP scope: Mobile + Desktop, device linking, recycling validation, finance module
  • ✅ Device linking: QR pairing + BLE auto-discovery, TLS, WebSocket
  • ✅ Image validation: ML-based (ONNX), Desktop processes, Mobile captures
  • ✅ BLE auto-discovery: WinRT advertising on Desktop, react-native-ble-plx on Mobile
  • ✅ Data storage: SQLite + JSON, fully local
  • ✅ Desktop redesigned as Zelara Core — local AI hub (2026-04-04)
  • ✅ Finance designed as standalone app — not a locked module (2026-04-04)
  • ✅ All mobile apps are standalone RN projects — no embed-then-extract pattern (2026-04-05)

In progress:

  • v0.6.0: Zelara Core redesign + @zelara/finance package + apps/finance-mobile/ + apps/testing-mobile/ standalone scaffolds

Core Concept

Zelara is a family of edge-first, privacy-first apps powered by a local AI hub.

  • Zelara Core (apps/desktop/, ai.zelara.desktop) — your personal AI hub; manages AI models, connects all Zelara apps
  • Zelara Finance (apps/finance-mobile/, ai.zelara.finance) — smart budgeting with AI receipt scanning and spend analytics
  • Zelara (apps/mobile/, ai.zelara.mobile) — lifestyle + skill tree; recycling validation, green causes, progression
  • Zelara Testing (apps/testing-mobile/, ai.zelara.testing) — developer diagnostic tool; validates device-to-device communication
  • Edge-first — all AI inference runs on your own machine, never a cloud server
  • No subscriptions — models download once to your machine, stored locally
  • Progressive unlocks — skill tree rewards cross-app engagement; no features gated for new users
  • Green initiatives — finance surplus → donation suggestions; recycling gamification

For Contributors

This is an AI-coordinated project. See CLAUDE.md in root for AI development guidelines.

All architectural decisions documented here require human approval before implementation.

Clone this wiki locally