WattPredict 瓦特预言
Hundreds of yuan on the electricity bill every month, and no idea where it went — most people "manage" their power usage by glancing at the meter, and saving energy is pure guesswork. WattPredict turns passive recording into proactive simulation: run the numbers before you act, translating kilowatt-hours into yuan, and get quantified, actionable conclusions like "raise the AC by 1°C, save ¥15 a month".
The same person might be a family renting in Shenzhen, a bubble-tea shop owner, and a small-factory investor — three completely different electricity scenarios, each with its own savings headroom. The scenario is the unit of value: residential tiered pricing, commercial time-of-use, and industrial two-part tariffs, all simulated in one platform.
Key Features
Bill simulation · WattEngine
Guangdong 2025 tariff spec: residential tiers (with summer/non-summer bands), commercial flat and time-of-use rates, industrial two-part tariffs (capacity/demand charges) — precise tier-by-tier, peak/flat/valley accounting; 22 appliance presets estimate monthly usage in one click and feed straight into the simulation.
Bill reconciliation · data flywheel
Enter a real bill amount and the system reverse-engineers your usage profile (monthly kWh + peak/valley split); deviation diagnosis, month-over-month insights, and accuracy trend tracking — the profile feeds back into simulation, so the model gets sharper over time. That's the moat.
Action recommendation engine
Tier-crossing early warnings (how many kWh to the next band, and what it costs), commercial TOU switch analysis (is time-of-use worth it?), and industrial capacity/demand declaration optimization — not just accounting, but what to do next.
Cross-platform
A Next.js 15 web dashboard (overview / scenarios / simulation / appliances / bill calendar heatmap) plus an Expo mobile app with four tabs (overview / simulation / appliances / bills) — simulate anywhere.
How It Works
Simulation results at a glance: baseline bill on top, per-scenario savings cards below, and the monthly reconciliation feedback strip at the bottom.
Real product screenshots to be added: the web dashboard, the bill calendar heatmap, and the mobile app.
Tech Stack
A monorepo with three moving parts: apps/web is the Next.js 15 dashboard, apps/mobile is the Expo app, and packages/wattengine is the standalone Go simulation engine (WattEngine) — a fully stateless compute service in a single binary with <100ms cold start, horizontally scalable; the web app calls it through Server Actions so the engine URL is never exposed to the client. Data lives in Supabase (PostgreSQL 15 + Auth + RLS); tariff rules are JSON-based and hot-updatable.
Stack: Next.js 15 (App Router) · React 19 · TypeScript strict · Tailwind CSS + shadcn/ui · TanStack Query · Zustand · React Native / Expo SDK 57 · Go 1.22+ (chi/v5) · Supabase (PostgreSQL 15 + Auth + RLS) · Vercel + Fly.io
Links
Roadmap & Status
v0.1: Guangdong province-wide · three meter types · simulation + bill reconciliation flywheel ✅ · v1: mobile app · bill calendar heatmap · seasonal tier bands ✅. Pricing: free tier ¥0 (1 scenario) · multi-scenario pack ¥8 one-time · Pro ¥18/year · Pro lifetime ¥30.
Planned next:
- Shanghai / Zhejiang tariff integration → nationwide coverage
- B-side SaaS: managed electricity bills for shops and business parks
- Long-term: VPP (virtual power plant) — aggregating load for demand response