Why there are two stages #
When you build an app with AI, it is better to test it in two steps:
- Local development/testing
- Production testing after deploy
This makes the process safer, faster, and easier.
Local development/testing #
Local testing is the best place for the first checks.
To view the feature locally, you need to enter fusebase dev start in the terminal, or simply type Open local server in the AI chat. This lets you test the feature in your browser right away. If something is broken, missing, or just not good enough, you can immediately tell the AI in chat. The AI can then fix the issue or improve the feature. In the latest versions, we have also enabled the CLI to automatically retrieve and fix logs, so you no longer need to write anything additional in the chat—the AI will detect and resolve issues on its own.
Why local testing is better first #
- It is faster
- You can see changes immediately
- It is easier to catch bugs early
- You can use logs and quick feedback during development
- The AI can help find and fix problems quickly
- You do not risk showing unfinished work to real users
In simple words: local testing is your safe workspace. This is where you experiment, debug, improve, and repeat.
Production testing after deploy #
Production testing happens after you deploy/publish the feature.
To deploy, you need to enter fusebase deploy in the terminal, or simply type Deploy or Publish in the AI chat. After deployment, FuseBase provides real feature links, and the app can be launched from the Apps section. The CLI guide also says deployment is how you make the latest changes available in production.
What production testing is for #
Production testing is useful when you want to:
- confirm that everything works in the real environment
- test the final version with real URLs
- check the full user experience
- prepare the feature for real clients or teammates
But production is not the best place for early debugging.
Best practice #
A good rule is:
Test locally first. Deploy only when you are confident everything works well.
Local testing is for:
- fast checks
- fixing bugs
- improving details
- reviewing logs and behavior
Production testing is for:
- final validation
- checking the live version
- sharing the app with others
Recommended workflow #
- Ask AI to build the feature
- Open the local server
- Check feature locally
- Fix issues (if any) with AI.
- Repeat until everything looks good
- Deploy/publish
- Test the live version one more time