Skip to main content
Version: 3.0

AI-Assisted Development

Steedos 3.0 is built around metadata-driven development. Objects, fields, permissions, pages, applications, and business logic can be represented as files inside a Steedos project. That makes Steedos a strong fit for AI-assisted development: the assistant can work with explicit project files, while developers keep control of requirements, review, testing, and release decisions.

AI should not replace business ownership. Treat it as a development partner that can draft metadata, explain platform concepts, generate package files, and help iterate on errors after you test the result.

  1. Describe the business requirement in natural language.
  2. Ask the AI assistant to produce a plan before changing files.
  3. Review the proposed objects, fields, permissions, pages, and automation.
  4. Let the assistant modify files under your Steedos project, usually inside steedos-packages/.
  5. Start the Steedos project and verify the result in the browser.
  6. Feed errors, screenshots, or observations back into the assistant and iterate.

What AI Can Help With

AI assistants are most useful for tasks that map to Steedos metadata:

AreaExamples
Data modelingObjects, fields, relationships, list views
User interfaceApplications, tabs, Amis pages, buttons
SecurityPermission sets, profiles, record access rules
Business logicTriggers, functions, workflow-related scripts
Data and reportingSeed data, analytics questions, dashboards
TestingBrowser test flows, issue reproduction, iteration notes

Human Review Still Matters

Before publishing AI-generated changes, review:

  • Whether the data model matches the real business process.
  • Whether permissions expose only the right records and fields.
  • Whether generated scripts are safe, maintainable, and scoped.
  • Whether the UI is clear enough for the target users.
  • Whether tests cover the workflow that will be used in production.

For Steedos-specific guidance, install and use Steedos Skills so your AI assistant can reason with the platform's conventions instead of generic low-code assumptions.