# Creative Agent: the complete guide

> Updated: 2026-10-30 · Applies to KiiKii
> HTML version: https://kiikii.cc/en/doc/create

## Quick answer

- Best for complete content sets: cover plus images, script plus storyboard plus final cut, talking head plus voiceover.
- The composer supports @ asset references, skills, and prompt optimization.
- Model, aspect ratio, and resolution can be left to the Agent or set manually.
- Generation shows thinking steps and elapsed time; failures report diagnostics and can be retried.
- Results can be downloaded, or you can copy the prompt, copy the link, or reference the result to keep creating.

## 1. What the Creation Agent is
The Creation Agent is KiiKii's conversational creation entry point. Describe your idea, script, or visual requirements clearly, and it understands the intent, breaks it into multiple tasks, selects skills and models, then produces images, video, audio, or copy in sequence. It solves four common problems: not being able to write prompts, needing many assets for one piece of content, expensive revisions, and assets scattered across tools that can be referenced repeatedly within a round.

## 2. Who it is for and what it can do
It suits creators and knowledge bloggers, e-commerce and brands, fiction and narrative content, local services and stores, and corporate marketing and training. Three typical workflows are "image first, then video", "script first, then storyboard", and "asset reuse" with uploaded people and products referenced by @; for a single image or a single 5-second clip, the matching studio is faster.

## 3. Interface layout and entry points
Open Agent in the top navigation; the left holds creation history and navigation, the center holds the conversation and multi-round generation records, and the right previews this round's results. New Conversation starts a separate topic, This Round's Creation Details shows which tasks were split out and which models and parameters were used, the Asset Panel holds this round's images, video, audio, and copy, and Copy Input reuses the previous prompt for a new topic.

## 4. Composer and parameters
The four buttons above the composer are the Agent's core actions: Agent (plan steps itself), @ Reference (name an uploaded asset and its role), Skills (assign a skill manually or let the Agent match one), and Optimize (rewrite a casual request into a fuller prompt). When writing a request, state the purpose, the subject, and the style, since the purpose determines aspect ratio and composition.

## 5. Reference assets and @ references
Uploaded assets do not take effect automatically—you must name them with @ and explain what they are for. Character assets keep the same look, product assets preserve packaging and logos, audio assets drive lip sync or voiceover, and video assets act as references for camera movement or pacing. Example: "Use the character from @Image 1 and the product from @Image 2 to generate a 9:16 e-commerce hero image, change the background to a light wood desk, and leave the price area clear on the right."

## 6. Using skills
A skill is a packaged creation workflow that includes the prompt structure, model selection, and parameters; once selected, the Agent follows it so results are more consistent. Use one when you produce the same type of content repeatedly, when there are clear industry requirements, or when you want to start from a proven template. Choose one primary skill at a time and add differences in text.

## 7. Thinking process and run status
The interface shows completed thinking, current generation, and elapsed time: a generating task shows the current step, a paused task usually awaits confirmation to avoid duplicate charges, and a failed task reports the original diagnostics and can be retried. You can leave the page during long tasks. Stop generation promptly when a result is off target; after a failure, read the diagnostics and add constraints rather than retrying the same prompt.

## 8. Results and actions
Each result card shows the model, aspect ratio, and quality actually used, and you can download the original file, copy the prompt as a template, copy the link, or reference the result as input for the next round. When unsatisfied, keep editing in the conversation—"keep the character, change the background to a rainy night street", "make the subtitles one size larger", "change it to 9:16"—instead of starting over.

## 9. Multi-turn creation in one conversation
The same conversation keeps its context. In the example, turn 1 switches to audio mode and generates a narration with an elderly male voice; turn 2 references the character image with "@Image 1 the character speaks a talking-head narration of this audio on a golf course"; turn 3 keeps refining scene, camera angle, or tone without reconfiguring. In later rounds you only need to say "this audio" or "this image", and when you switch topics you click New Conversation so projects do not interfere.

## 10. Work showcase and demos
All works come from the same entry point: a vertical social media cover for a Xining travel guide, a 9:16 character image of a woman in historical dress dancing in a forest, a 4:3 photorealistic desk scene, plus two 15-second films (a wuxia showdown on a rainy night bridge and a cloud whale gliding over a coastal city). Before publishing, verify dialogue, likenesses, and asset rights, and label AI-generated content as the platform requires.

## FAQ

### How is the Creation Agent different from the image and video studios?
A studio is single-shot generation: pick a model, set parameters, get one image or one video. The Agent is continuous, conversational creation: you state the goal, and it breaks the work into tasks, selects skills and models, writes a script or storyboard first when needed, then produces results in sequence.
### Do I still need to choose models and parameters myself?
Not necessarily. By default the Agent selects the model, aspect ratio, and resolution based on the task, and the result card shows what was used. If you have specific requirements, set them manually below the composer.
### How do I use reference assets?
Click Add Assets to upload an image, video, or audio file, then type @ to select the asset and state its role, for example "Use the character from @Image 1 to generate a 9:16 talking-head video in a café". The asset acts as a reference rather than being pasted in.
### What if a task is paused or fails?
A failed task reports the original diagnostics and can be retried directly. A paused state usually avoids duplicate charges—confirm the upstream result and continue. Results already generated are not lost when a later task fails.
### Where are generated works saved?
Each round's results go into this round's creation assets, visible in the asset panel on the right of the conversation and filterable by source in My Assets. Result cards support download, copying the prompt, copying the link, and referencing the result.
### Can I use the output commercially?
Make sure you hold the appropriate rights to any uploaded reference assets, likenesses, music, and copy, and comply with platform rules and local law. Some platforms require AI-generated content to be labeled, so follow the target platform's requirements.