Working with your AI assistant
You build Cloudrizz apps by chatting. The assistant does the work through the Cloudrizz tools — so the quality of what you get tracks the clarity of what you ask. A few habits make a big difference.
Describe the outcome, not the code
Say what the app should do and who it's for. “A booking page where clients pick a time slot and pay a deposit” is better than naming files or functions. The assistant chooses the structure; you steer the result.
Iterate in small steps
Ship something tiny, look at it, then refine. Small changes are easier for the assistant to get right and easier for you to verify. After each change it can deploy_app so you see the live result before moving on.
- “Add a contact form.” → review → “Email me the submissions.”
- Prefer a sequence of clear asks over one giant prompt.
Use branches for big or risky changes
For a substantial feature or a rewrite, ask the assistant to work on a branch and give you a preview URL. You can try it in isolation — against a throwaway database — before it touches production. When you're happy, it opens a pull request to merge. See Branches, PRs & AI merges.
When something breaks, say so plainly
Paste the error or describe what you saw (“the page is blank after I click Save”). Ask the assistant to check the deployment logs — it can read them and usually pinpoint the cause. See Troubleshooting.
Lean on what's built in
You rarely need third-party services — ask for the platform features by name and the assistant wires them up:
- “Add user accounts with email sign-in” → auth
- “Store uploaded photos” → storage
- “Email a receipt” → email
- “Charge a subscription” → payments
- “Summarize this with AI” → AI
- “Send a digest every morning” → scheduled jobs