Attach AI text, image, and voice generation to Payload CMS fields

What it does

This plugin embeds AI generation into the Payload CMS admin interface, adding buttons for composing, proofreading, translating, and rephrasing text directly inside richText and other fields. It also hooks into upload fields to generate images via DALL-E and GPT-Image-1, plus voice audio through ElevenLabs and OpenAI TTS. Configuration happens at the field and collection level, with access controls to limit who can trigger prompts.

The interesting bit

Rather than forcing a single vendor, it lets you bring your own keys for OpenAI, Anthropic, Google Gemini, or MiniMax, and even define field-level prompts so the AI behaves differently depending on which input it is editing. That turns a generic chatbot into a context-aware content assistant tied to your actual data model.

Key highlights

• Supports multiple AI backends for text (OpenAI, Anthropic, Google, MiniMax), image (OpenAI, Google), and voice (ElevenLabs, OpenAI).
• Field-level customization via prompt overrides and a built-in prompt editor.
• Access control hooks to restrict AI features by user role.
• i18n support and custom UI component extension points.
• Includes beta features for proofreading and rephrasing, with expansion and summarization marked as coming soon.

Caveats

• The plugin is explicitly in beta and tested against a specific Payload version (v3.38.0), so expect churn.
• Several headline features—expand, summarize, simplify, document analyzer, fact checking, and automated workflows—are listed as “Coming Soon” and not yet implemented.
• Custom fields do not automatically inherit AI capabilities and require manual component configuration.

Verdict

Content teams already using Payload CMS who want to stop tab-switching to ChatGPT will find this a natural fit. If you are not on Payload v3 or prefer a completely decoupled AI pipeline, this is not your tool.