AI Editor
Definition
A CE.SDK configuration that combines standard design and video editing with generative AI capabilities (image, video, text, and voice creation) surfaced directly inside the editing interface. Unlike standalone AI generation tools, the AI Editor embeds generation into a full composition surface: users can prompt, generate, refine, and export in a single session without switching applications or workflows.
All AI-generated content lands in the scene as standard CE.SDK blocks: editable, movable, and exportable like any element added manually. There is no distinction between AI-generated and hand-crafted content once it enters the scene; both pass through the same rendering pipeline.
Model-Agnostic Architecture
The AI Editor does not bundle or require specific AI providers. Each generation capability is implemented as a CE.SDK plugin, and the models or APIs used are selected and configured by the product team embedding the editor.
This means the same editor shell can run OpenAI image generation for one product, fal.ai for another, and ElevenLabs voice synthesis for a third, without changes to the underlying UI structure.
Licensing note: AI plugins are included in the existing CE.SDK license. The underlying AI model APIs require separate licensing from the respective providers.
AI Generation Capabilities
Image Generation (fal.ai, OpenAI)
Text-to-image and image-to-image, producing raster or vector output in styles from photorealistic to illustrated. Supported providers include fal.ai and OpenAI.
Video Generation (Model-agnostic)
Prompt-to-video clips inserted directly onto the timeline or canvas as standard video blocks.
Text Generation (Claude, GPT)
AI-assisted copywriting for headings, body text, and captions inside text blocks.
Voice & Audio Generation (ElevenLabs)
Text-to-speech voiceovers and audio generation inserted as audio blocks on the timeline.
Background Removal (Client-side)
Client-side ML model. No external API call required. Photos never leave the device.
Style Transfer & Image Manipulation (Model-agnostic)
Variant generation, object add/remove, and style transfer powered by configurable third-party models.
Typical Use Cases
AI-Native Ad Creation Platforms
Closes the gap between raw generation and a campaign-ready asset. Generation and composition happen in the same surface, so users never need to export a draft into a separate editor to finish it.
Generative Design Tools
AI produces a starting point (a background, a product image, a layout suggestion) and the user refines it before export. Generation outputs land directly as CE.SDK blocks, ready to reposition, resize, and combine with manually created elements.
High-Frequency Content Production
AI drafts volume (ad variants, social posts, product imagery) and the editor surfaces a review and adjustment step before assets go live. The model-agnostic architecture means the AI provider can be swapped or upgraded without rebuilding the interface.
AI-Assisted Template Population
Generation tools fill placeholder content in a pre-built template (writing localized captions, generating product imagery, or suggesting headline copy), and the editor presents the populated result for a final human review before export or distribution.
Embedded Voiceover Production
Teams produce narrated videos entirely within one tool: generate a voiceover, place it on the timeline, sync it with video clips and text overlays, and export, without routing audio work through a separate application.
Web (Vanilla JS/TS, React, Vue, Angular, Svelte, SvelteKit, Next.js, Nuxt.js), Electron. Mobile availability is planned.
Links
Documentation and references for this concept.
Related Terms
Starter Kit
A pre-configured, production-ready CE.SDK setup for a specific use case, ready to embed, customize, and ship, providing …
Plugin
A modular extension registered at editor initialization via cesdk.addPlugin() that adds custom functionality to CE.SDK (…
Design Editor
CE.SDK's full-featured design creation Starter Kit, exposing the complete toolset: multi-page canvas, typography control…