How JENVU Is Built

A modern, edge-rendered, type-safe stack tuned for sub-second voice latency and reliable signal generation.

Frontend

  • React 19 with TanStack Start (file-based routing, SSR-ready).
  • Vite 7 + Tailwind v4 for instant builds and design tokens.
  • Framer Motion for entrance animations and the floating orb.
  • lightweight-charts for institutional-grade dual-timeframe rendering.

Voice Layer

The Web Speech API powers low-latency speech-to-text and text-to-speech directly in the browser. Word-boundary events drive the orb's pulsing animation so the agent feels alive while it speaks.

Server Functions

Market data fetching, news aggregation, and LLM orchestration run as TanStack server functions on a serverless edge runtime. Each XAU cross-pair routes to OANDA and Yahoo Finance bullion feeds with automatic fallbacks for resilience.

Backend

  • Managed cloud backend for authentication and persistence.
  • Row-level security on every table; roles isolated in a dedicated table.
  • Edge functions for webhooks and scheduled tasks only.

Principles

  • Type safety from route params to server-function payloads.
  • Schema-bound LLM outputs — the UI never trusts free-form text.
  • Honest defaults — when uncertainty is high, the agent stands aside.
  • Performance budget: speak in under one second.

Last updated · July 2026