SupportPilot AI
A production-style AI customer support and WeCom CRM dashboard with CRM, inbox, AI reply suggestions, knowledge base, analytics, templates, settings, and demo login.
Business value
Helps support teams organize customers, respond faster with reviewed AI suggestions, and understand performance through analytics.
Key features
overview
This is a live portfolio project that shows how AI can be placed inside a real support workflow instead of a simple chatbot.
problem
Many small teams have scattered customer messages, slow replies, and no shared knowledge base or visibility into support performance.
solution
Build a single workspace that combines CRM, conversation inbox, knowledge base content, AI reply suggestions, templates, and analytics.
key features
Demo login / Customer CRM / Three-column inbox / AI reply suggestions / Knowledge base / Analytics charts / Templates / Settings / Chinese/English UI
tech stack
Next.js / React / TypeScript / Ant Design / Tailwind CSS / Supabase / Prisma / Vercel AI SDK / DeepSeek/Qwen API / TanStack Query / Zustand
architecture
Next.js App Router handles pages and APIs, Prisma schema prepares for Supabase PostgreSQL, and demo mode uses typed mock data plus an AI fallback.
AI-assisted workflow
AI generates suggestions from customer messages, customer profiles, and knowledge base context, but the agent must review, edit, and send manually.
challenges and what I learned
The main learning goals were enterprise dashboard architecture, AI safety boundaries, Ant Design tables, bilingual UX, and maintainable data structures.
screenshots
Available through the live demo.