Introduction

CryptoDesk AI is a non-custodial desktop trading co-pilot built around a careful principle: users should gain more clarity, not less, when they use AI and trading automation concepts.

Security-first posture

The current alpha is intentionally conservative. Live trading remains gated while simulation, documentation, and release hardening continue to mature.

What is CryptoDesk AI?

CryptoDesk AI combines real-time market visibility, read-only portfolio preview, simulation trading, and explainable AI recommendations inside an Electron desktop application. The goal is not to act as a black-box trading bot. The goal is to help users make better-informed decisions while preserving strict local-control boundaries.

Core Principles

Non-Custodial

Credentials are meant to stay on the user machine. The renderer does not receive direct secret access, and privileged work flows through preload and main-process validation.

Simulation-First

New users start with a virtual balance and safe paper-trading flows. The simulation loop exists to make learning and verification practical before any future live execution path is considered.

Explainable AI

Signals are advisory-only. Instead of hiding behind a confidence number alone, the product shows reasoning, filtering, and downgrade behavior such as turning a risky signal into hold.

Bilingual Experience

The renderer supports both English and Arabic, including RTL-aware behavior for Arabic where appropriate.

Who Is It For?

  • Traders who want a desktop-first workflow instead of a browser-only stack
  • Learners who want simulation and risk context before touching real exchange operations
  • Reviewers and contributors who care about architecture, security boundaries, and release discipline

Technology Foundation

  • Desktop shell: Electron 28
  • Renderer: React 18 and TypeScript 5 in strict mode
  • Exchange layer: CCXT
  • Storage: SQLite and local persistence flows
  • AI runtime: TensorFlow.js and ONNX Runtime
  • Validation: Zod-backed IPC contracts

The result is a product shaped as much by trust boundaries and operational clarity as by visible features.

Edit this page