AI Engineer for Generated Apps

Applies to: App Generation projects that use the SaaS Engine template

The AI Engineer changes and troubleshoots a generated application inside its hosted development workspace.

It is different from the AI Architect. The AI Architect gathers requirements before project creation.

What the AI Engineer can do

The AI Engineer can help with these tasks:

  • implement application workflows.
  • change pages, layouts, and navigation.
  • update frontend and backend source.
  • add validated API behavior.
  • create database models and migrations.
  • inspect errors and logs.
  • run project checks.
  • explain completed changes.

The result depends on the request, current source, application state, model, and available tools.

Initial application work

After creation, AppWizzy sends the approved requirements to the SaaS Engine workspace. The AI Engineer implements an initial useful application workflow.

The initial work extends the SaaS foundation. It can include public pages, navigation, forms, data flows, and supporting API changes.

Review this first result as an initial iteration. Test it before you request unrelated changes.

Request a change

  1. Open Edit via Chat.
  2. Describe one required outcome.
  3. Name the affected user and workflow.
  4. Provide acceptance criteria.
  5. State important limits.
  6. Attach a relevant image when useful.
  7. Send the request.
  8. Wait for the operation to finish.
  9. Reload and test the preview.
  10. Save a version after approval.

Example request:

Add invoice approval. Managers can approve pending invoices. Staff can view status but cannot approve an invoice.

This request identifies the workflow, roles, action, and access rule.

Use images and preview context

You can attach screenshots or visual references to a request. The AI Engineer receives the selected attachments with the message.

Use the preview screenshot control when the visible application state explains the issue. Add text that identifies the incorrect result.

Do not place credentials in prompts or screenshots. Store secrets through the supported project configuration.

Select the model and reasoning level

SaaS Engine projects can show model and reasoning controls in the AI Engineer header. Available options can change.

Use a faster option for small, clear edits. Use more reasoning for complex workflows, migrations, or difficult failures.

Model choice can change AI usage cost and completion time.

Stop an active request

Use Stop when the request is incorrect or no longer required. A stopped operation can leave partial workspace changes.

Review changed files after a stop. Reset unsaved work or request a correction before you save a version.

Review every result

Test the affected workflow, roles, validation, mobile layout, and stored data. Review logs when the visible result is incomplete.

Save a version only after the result works. A saved source version does not include database records or the complete VM state.

Read Version Control and Rollback for recovery controls.

AI usage and billing

The AI Engineer uses managed AI access through the AppWizzy proxy. AppWizzy can route requests to different model providers.

AppWizzy charges the related credits to your account. Review the billing page for model usage and credit transactions.

Personal AI login and BYOK depend on the selected workspace template. They are not controls in the SaaS Engine AI Engineer chat.

Read AI Access Modes and Pricing, Plans, and Credits.