What You Get

Supabase replaces 3-4 separate services with one platform:

• Database: PostgreSQL with auto-generated REST and GraphQL APIs from your schema
• Auth: Email, OAuth (Google, Apple, GitHub), magic links, phone OTP — managed
• Storage: File uploads with row-level access control
• Edge Functions: Server-side logic on Deno
• Real-time: Subscribe to database changes as they happen

Row-level security (RLS) is the feature that matters most. Write one policy — "users can only read their own data" — and every API call respects it. No middleware. No custom auth checks in every endpoint. One rule, enforced at the database level.

The Economics

Supabase's free tier:

• 500MB database
• 1GB file storage
• 50,000 monthly active users
• Unlimited API requests

I run 4 production apps on Supabase. Monthly cost: $0. Not a trial. Not a promotion. The free tier handles indie-scale apps indefinitely.

When you outgrow it, Pro is $25/month. Compare that to running separate services: managed PostgreSQL ($15-50/month), auth provider ($0-30/month), file storage on S3 ($5-20/month), real-time service ($10-50/month). Supabase consolidates all of that.

For indie developers, fewer services means fewer things to configure, monitor, and debug. That simplification matters more than any individual feature.

No Lock-in

Supabase runs on PostgreSQL. Standard SQL. Standard tooling.

If you need to leave, you export with pg_dump and import anywhere that runs Postgres. No proprietary query language to rewrite. No data format to convert. No migration project that takes months.

Firebase uses a proprietary NoSQL database. Migrating away from Firebase means rewriting every query in your application. I have watched teams spend months on Firebase migrations. With Supabase, it is a database export.

Ownership matters. You should own your code, your data, and your infrastructure choices. Supabase respects that principle.

Why AI Agents Work Well with Supabase

PostgreSQL is the most documented, most trained-on database in the AI era. When an AI agent needs to create a table, it writes standard SQL. When it needs a migration, it generates a .sql file with CREATE TABLE and ALTER TABLE statements.

The agent does not need to learn a proprietary SDK. It does not need to understand nested document structures or platform-specific indexing rules. It writes SQL. SQL is universal. AI handles it reliably.

This is a practical advantage. AI-generated database code for PostgreSQL is more reliable than AI-generated code for NoSQL databases, simply because there is vastly more training data for SQL.