Choisir une base de données par consigne

There are many kinds of databases, and they are good at different jobs. The two big families you'll meet most are relational (like PostgreSQL or SQLite) and document (like MongoDB).

When vibe coding, you don't have to memorize them all. You describe your needs and let the AI recommend.

A relational database stores data in tables with rows and columns, like a spreadsheet, and lets tables link to each other.

It shines when your data is structured and connected, for example users who each own many orders.

A document database stores data as flexible JSON-like documents instead of strict tables.

It is handy when each record can have a different shape, or when you are still figuring out your structure. The trade-off is weaker guarantees about how data connects.

Instead of guessing the right tool, describe your app and ask for a recommendation. Good prompts include what you store, the rough scale, and your priorities.

The AI can match those needs to a database far faster than browsing comparison articles.

For many new projects, SQLite is a great first choice. It is a relational database that lives in a single file, needs no server, and is easy to set up.

You can always move to a bigger database later, and a good AI prompt can ask for an easy upgrade path.

You can run a database yourself, or use a managed service like Supabase or PlanetScale that hosts it for you.

Managed services save setup time and handle backups, which is great for solo vibe coders. Mention your preference so the AI tailors its setup steps.

Your database choice should play nicely with the rest of your stack. A Next.js app, a Python script, and a mobile app each have smooth and rough options.

Tell the AI your framework so it recommends a database with good support and clear setup instructions.

Some databases are free at small scale, then charge as you grow. A surprise bill is no fun.

Ask the AI about pricing tiers and free limits before committing, so your choice fits both your project and your budget.

If you ask "which database is best?" you'll get a generic answer. Best depends on your situation.

The more context you give, the sharper the recommendation. Treat the AI like a consultant who needs the full brief, not a magic oracle.

Once you've chosen, ask the AI to record the decision and reasoning in your project, often in a README or a notes file.

This keeps future you (and future prompts) consistent, so you don't accidentally drift to a different database mid-build.

Good news: early database choices are rarely permanent. If you keep your data access in one place, swapping databases later is far easier.

You can even ask the AI to structure your code so the database is easy to replace down the road.

Test your understanding of choosing a database by prompt.

You met relational and document databases and learned to choose by describing your app, not by dictating a tool. SQLite is a friendly starting point, and managed services save setup work.

Always share stack, scale, and budget in your prompt, and record the decision. Next, you'll model your tables with AI.