Backlit vs Supabase and Firebase: when you just want hosting + storage
Supabase and Firebase give you a whole backend to build on. Backlit gives you a finished home for a single-page app. Here's how to tell which one your project actually needs.
When an AI-built app needs a login and saved data, the advice you’ll often hear is “use Supabase” or “use Firebase.” They’re excellent platforms. But for a single-page app you just want to share, they can be far more than you need. Here’s a clear comparison to help you tell the difference.
The short answer
Use Supabase or Firebase if you’re building a real, evolving application and you (or a developer) will design and maintain a backend, a database schema, security rules, queries, the lot. Use Backlit if you have a finished single-page app and just want it hosted with sign-in and storage already attached, set up by your assistant, with nothing to maintain. The first is a backend you build; the second is a home that comes finished.
Side by side
| Supabase / Firebase | Backlit | |
|---|---|---|
| What it is | A backend toolkit you build on | A finished host for single-page apps |
| Hosting included | Partial (you often add a host) | Yes, the app lives here |
| Auth | Powerful, you configure & integrate | Built in, magic link, Google, Microsoft, allowlist |
| Database | Full database you design & secure | Key-value storage: shared, per-user, capture |
| Who it’s for | Developers building applications | Anyone sharing an AI-built single-page app |
| Setup effort | Medium to high, ongoing | Very low, one-time |
| Set up by your assistant | No | Yes, via MCP, in the chat |
Where Supabase and Firebase shine
These are real backends. Supabase gives you a full Postgres database, auth, storage, and APIs; Firebase gives you a real-time database, auth, functions, and a deep Google-cloud ecosystem. If you’re building something that will grow (many features, complex relationships in the data, fine-grained permissions, server-side logic), they’re a strong foundation, and they scale a long way.
The cost is that they hand you capability, not a finished thing. You design the schema. You write the security rules, and getting those wrong is a real risk: it’s a common way apps built with AI get hacked (how to keep an app you built with AI from getting hacked). You integrate the SDKs, wire auth into your front end, and usually still pick a host for the actual app. For a developer building an application, that’s the job. For someone who built a single-page tool in a chat, it’s a backend project bolted onto a ten-minute app, and a backend you now have to maintain.
Where Backlit shines
Backlit isn’t a backend you build on; it’s the place a finished single-page app lives, with the backend parts already wired:
- Hosting + auth + storage as one thing. You don’t assemble three services. The app, its sign-in, and its data come together.
- No schema, no rules to get wrong. Storage is simple key-value with three surfaces (shared data, per-user silos, and write-only captures), reachable from your HTML via
window.backlit.*. There’s no security-rules language to learn or misconfigure; per-user isolation is built in. - Your assistant sets it up. Connect Backlit over MCP and ask Claude or ChatGPT to publish. No SDK integration steps. The same connection lets you query or export your app’s data straight from Claude later, instead of writing queries yourself (how to get your app’s data back out).
- Nothing to maintain. No database to manage, patch, or scale. It scales to zero when idle, which is part of why it’s a couple of dollars a month.
The trade-off is deliberate scope: Backlit’s storage is key-value for single-page apps, not a relational database with complex queries. If you need joins, transactions across many tables, and rich server-side logic, that’s a real backend’s job. Reach for Supabase or Firebase.
A useful way to decide
Ask what kind of data thinking your app needs:
- “Save this person’s stuff, save this shared thing, collect these submissions.” → That’s Backlit. Three calls, done.
- “Model these five related entities, query across them, enforce these per-role permissions, run this logic server-side.” → That’s a database, and you want Supabase or Firebase.
Most apps people build with AI and want to share are firmly in the first camp. They just feel like they need a database because that’s the only path anyone mentioned. If your data needs fit in a few keys, you can skip the backend entirely and still have real, persistent, per-user storage. Read the FAQ → or see how it works →.
Built an app with Claude or ChatGPT? Get early access to Backlit and share it in seconds.