AI-assisted model design
Create and refine PostgreSQL Tables, fields, relationships, labels, validation, and editor guidance with OpenAI or Gemini, then review the complete proposal before applying it.
What works now, what is being validated during public beta, and what comes next for the AI-driven, self-hosted headless CMS.
GoaCMS can connect to PostgreSQL, discover supported existing structures, and add CMS management metadata without claiming ownership of the underlying Tables or rewriting their rows.
Review the detected structure first, then choose the Tables that should become managed Lists. GoaCMS enables only the operations that the detected keys, defaults, generated columns, and constraints can support safely.
These capabilities are implemented today. Public-beta feedback is helping refine their operating boundaries and documentation.
Create and refine PostgreSQL Tables, fields, relationships, labels, validation, and editor guidance with OpenAI or Gemini, then review the complete proposal before applying it.
Every managed List receives authenticated, permission-aware item endpoints with field selection, filters, search, sorting, pagination, relation expansion, and optional public reads.
Configure roles, field boundaries, row filters, create presets, versions, activity, comments, assets, import and export, live previews, and signed webhook Flows.
The existing-database workflow has focused unit coverage. The following work will add full-system evidence before GoaCMS describes it as production-validated across diverse legacy schemas.
Automate discovery, reviewed management, generated API create/read/update/delete behavior, and row preservation against real PostgreSQL fixtures rather than only isolated logic tests.
Expand fixtures for multiple schemas, foreign keys, enums, arrays, generated fields, unusual defaults, large identifiers, views, and unsupported types with actionable diagnostics.
Measure discovery and API performance on large Tables, then verify that external schema changes remain visible, reviewable, and non-destructive across supported PostgreSQL releases.
The next AI capabilities move beyond model setup while retaining explicit permissions, previews, and human control.
Create, update, translate, classify, and prepare content for publishing through permission-aware AI actions with a reviewable activity trail.
Compose pages from managed content and propose role, field, and row-access changes for an administrator to review before they take effect.
Give approved AI agents a documented, scoped interface for discovering models and operating content without bypassing GoaCMS validation or permissions.
Join the public beta, test GoaCMS with your PostgreSQL model, and use the feedback button on any page to share the database patterns and AI workflows you need.
Tell us what worked, what did not, or what you need next.