Invoicey
A production-focused invoicing app built with Next.js App Router, Firebase Auth, and MongoDB. Helps users create, edit, manage, preview, and export professional invoices with AI-assisted drafting.
Timeline
1 month
Role
Full Stack
Team
Solo
Status
CompletedTechnology Stack
Key Challenges
- Firebase Admin SDK token verification in API routes
- Multi-format invoice export (PDF, HTML, CSV, JSON)
- AI-assisted invoice drafting with follow-up clarification
- Line item management with live preview
Key Learnings
- Firebase client and Admin SDK split auth architecture
- MongoDB Mongoose modeling for invoicing workflows
- Building multi-stage Docker image with Bun runtime
Overview
Invoicey is a production-focused invoicing app designed for freelancers, founders, and small teams that need a simple but polished billing workflow. Built with Next.js App Router, Firebase Auth, and MongoDB.
Core Features
- Google Sign-In via Firebase client SDK
- Invoice CRUD with Firebase ID token verification on every API route
- MongoDB persistence with Mongoose models and soft-delete support
- AI-assisted drafting — describe an invoice in plain English and Gemini fills in the fields with follow-up clarification questions
- Multi-format export — Print/PDF, HTML, CSV, and JSON
- Live preview while editing with line item management, date pickers, and currency selectors
- Persistent dark/light mode initialized before hydration to avoid flash
Tech Stack
| Layer | Technology |
|---|---|
| Framework | Next.js 16 (App Router) |
| Language | TypeScript |
| Auth | Firebase Client + Admin SDK |
| Database | MongoDB + Mongoose |
| UI | shadcn/ui + Aceternity patterns |
| AI | Gemini via /api/ai/invoice-assistant |
| Runtime | Bun |
| Deployment | Vercel |
Architecture
Authentication uses a split model: the Firebase client SDK handles Google sign-in in the browser, and the Firebase Admin SDK verifies ID tokens server-side on every API request. The session token returned after verification is used to scope all invoice reads and writes by userId in MongoDB.
The AI assistant accepts a natural language prompt plus the current form draft, sends both to Gemini, and returns a structured patch payload along with clarification questions if any fields are ambiguous.
Export Pipeline
Invoicey supports four export formats from a single invoice record:
- Print/PDF — generates HTML and triggers
window.print() - HTML — downloads a self-contained HTML blob
- CSV — flattens line items into comma-separated format
- JSON — raw invoice record for programmatic use
What I Learned
Building Invoicey deepened my understanding of the Firebase Admin SDK token verification flow and how to structure MongoDB schemas for financial records with soft-delete and settle-to-paid lifecycle transitions. The multi-stage Bun Dockerfile was a first for me and significantly reduced the production image size.