- Published on
Quickly generate Postgres database in a single command
How I thought I could let AI do all the work but it turns out it can only do so much
- Authors
- Name
- Nico Prananta
- Follow me on Bluesky
I was curious if it's possible to create a small app entirely by AI. After a day or so, with the help of Claude Sonnet 3.5 and GPT-4 via Raycast, I finally created a Node.js CLI tool called create-postgresql that can quickly generate a Postgres database in a single command. You can use it for FREE and without any registration, but the server deletes all the generated databases every day at midnight UTC.
Image
The whole app actually consists of several parts:
- A Node.js CLI tool that connects to the server and creates the database.
- An Express server that receives the request from the CLI tool and creates the database.
- A static web page to verify the request before creating the database.
- A PostgreSQL instance that is used to store the databases.
- A PgBouncer instance that is used to pool the PostgreSQL instance.
I wanted all the server parts including the PostgreSQL and PgBouncer to be deployable easily using a docker compose file. After a lot of back and forth with the AI, I finally got it working. The whole server parts are deployed using Coolify to my server in Hetzner. You can host them yourself if you want.
You can quickly give it a try by running the following command:
npx create-postgresql
To prevent abuse, the CLI will open up a static web page where you can verify the request before creating the database. The submission is rate-limited and protected by Google reCAPTCHA v3.
If you want to use your own server, you can specify the URL via the command line:
npx create-postgresql --api-url https://api.example.com --frontend-url https://app.example.com
Closing Thoughts
Overall, the experience confirms my belief that at the current level, AI is a great tool to create software only if you have the software development skills. For example, I still need to tell the AI if some code is not right or not efficient. Moreover, the code it generates doesn't work most of the time. I need to tell it the error message, and it will attempt to fix the code. There were times when I had to manually fix the code it generated because it just couldn't figure out the problem.
AI will not replace software developer jobs in the near future. On the other hand, it will make software developers more productive.
Are you working in a team environment and your pull request process slows your team down? Then you have to grab a copy of my book, Pull Request Best Practices.
Image
