I'm a solo developer who built UnitHub, an AI-powered property management app for self-managing landlords. Here's the tech stack and what I
learned.
The Stack
| Layer | Tech |
|-------|------|
| Framework | Next.js 16 (App Router) |
| Database | Supabase (PostgreSQL + RLS) |
| Auth | Supabase Auth + Google OAuth |
| Payments | Stripe (subscriptions + Connect) |
| AI | OpenAI GPT-4, ElevenLabs Voice |
| Email | Resend |
| Hosting | Vercel |
| Analytics | PostHog |
AI Features I Built
1. Tenant Screening
GPT-4 analyzes rental applications and returns a risk score (0-100) with detailed reasoning. Takes 30 seconds vs days of manual review.
2. Voice Assistant
Using ElevenLabs, landlords can speak commands like "Add a maintenance request for unit 3B - broken dishwasher" and the app pre-fills forms.
3. Document Extraction
Upload a lease PDF → AI extracts tenant names, rent amount, dates, deposit automatically using GPT-3.5.
4. Smart Categorization
Maintenance requests auto-categorize (plumbing, electrical, HVAC) with priority detection.
Key Learnings
1. Supabase RLS is powerful but tricky
Row Level Security policies ensure tenants only see their own data. Debugging RLS issues was my biggest time sink.
2. Next.js 16 Server Components changed everything
Data fetching in components without useEffect spaghetti. Game changer for complex dashboards.
3. AI costs add up fast
Started with GPT-4 everywhere, switched to GPT-3.5-turbo for simple tasks. 90% cost reduction.
4. Ship early, iterate fast
Launched MVP in 8 weeks. User feedback shaped the roadmap more than my assumptions.
What's Next
- Multi-language support (i18n with next-intl)
- Calendar view for rent/maintenance
- Mobile app (maybe React Native?)
Try It
If you manage rental properties (or know someone who does), check out UnitHub. Free tier available.
Questions about the stack? Drop a comment
Top comments (0)