AI Workstation Credits MVP
A full-stack AI SaaS billing-loop demo with auth, user credits, backend-only AI calls, usage logs, admin recharge, and secure role checks.
业务价值
Shows senior-review-ready product thinking: failed AI calls do not deduct credits, successful calls deduct atomically, and admins can recharge users safely.
核心功能
overview
This MVP demonstrates the core business loop behind a real AI SaaS product rather than only a chat UI.
problem
AI SaaS products need to protect provider keys, control credit balances, avoid charging failed requests, and keep a reliable audit trail.
solution
The app routes all AI calls through a backend API, checks balance before provider calls, deducts credits only after a successful response, logs every completed attempt, and gives admins a protected recharge workflow.
key features
Supabase auth / 1000 default credits / Backend AI route / Atomic credit spend / Usage logs / Admin recharge / RLS security / Vercel deployment
tech stack
Next.js App Router / TypeScript / Tailwind CSS / Supabase / PostgreSQL RLS / OpenAI SDK / DeepSeek API / Zod / Vercel
architecture
Next.js App Router handles pages and route handlers. Supabase Auth manages users. PostgreSQL tables store profiles, usage logs, and immutable credit transactions. Service-role APIs perform trusted cross-user operations.
AI-assisted workflow
The AI provider is called only from the server through the OpenAI-compatible SDK. DeepSeek is configured by environment variables, so the provider can be changed without application code changes.
challenges and what I learned
The key learning was designing the secure billing boundary: RLS for user reads, service-role APIs for trusted writes, database functions for concurrency-safe credit changes, and clear documentation for technical review.
screenshots
Available through the live demo.