AI agent for images and video
AI Media Agent: Chat your idea into finished images and videos
Describe what you want in plain words. The agent picks the right model, writes the prompts, plans every scene and shows you the plan before anything runs. Approve it, and it generates the images, videos, avatars and voice-overs for you, then keeps refining with you in the same chat.
No prompt engineering
Say what you want in your own words, in any language. The agent chooses the model, writes a detailed prompt for each shot and tells you why.
A plan before any credits are spent
Every request becomes an editable plan with the model, length, references and estimated credits for each step. Nothing runs until you approve. Auto mode condenses it to one short summary.
Consistent characters across scenes
A character sheet anchors the identity and each scene references the one before, so faces, outfits and style stay the same from the first shot to the last.
The leading models in one chat
Seedance 2.5, Wan 3.0, Kling 3.0 and Gemini Omni for video; GPT Image 2.5, Nano Banana and Seedream 5.0 for images. Let the agent choose, or pick one yourself.
Your own references
Attach images, video clips and audio, or paste a YouTube link, and the agent plans a new piece in the same style, pacing and camera work.
Skills at a slash
Type / to pick a skill, such as a three-view character sheet, and the agent writes every scene with that skill's rules.
Memory and saved chats
Tell it once about your brand, format and tone, and memory carries it into every new chat. Save a conversation and reload it later with its media.
Straight into editing
Results save to My Stuff automatically. Send clips to the Video Editor, edit them or upscale them to 4K, or reuse any result as the next reference.
Multi-scene stories and short dramas
Ask for a story up to 60 seconds long. The agent writes the storyboard, keeps the character consistent with a character sheet, chains each scene to the one before and offers to combine them into one video.
Product ads and e-commerce videos
Upload a product photo and ask for an ad. The agent refines the shot first, then animates it with studio lighting and smooth camera moves, in the aspect ratio your channel needs.
Edit an existing video with a prompt
Swap a character, change the background or replace an object in a clip you already have. Pick a preset or describe the change, and choose from five video-edit models.
Layered image editing
Interactive Edit opens your image on a canvas with layers. Move, sketch and annotate, then regenerate exactly what you marked, or split a picture into separate editable layers.
Camera moves you direct
Pick a preset camera move, or draw your own path in a 3D preview for MiniMax H3 Max Camera Control. Seedance, Wan and H3 follow the move from the prompt.
Motion transfer
Give it a reference video of a dance or an action and a character image, and the motion is carried onto your character.
Talking avatars and voice-overs
Turn a portrait and a voice track into a talking presenter with OmniHuman or Kling Avatar, or generate natural voice-overs from your script.
Frequently asked questions
What is the AI Media Agent?
A chat-based AI creative assistant on AI9app. You describe the image or video you want; it chooses the model, writes the prompts, plans the steps and generates the result once you approve the plan.
Do I need to know how to write prompts?
No. Write the way you would brief a person, in English, Chinese, Japanese, Korean or another language. The agent turns it into detailed model prompts, and you can still open and edit each prompt in the plan.
How much does it cost?
Chatting with the agent uses your AI usage allowance, and each generation is billed in credits. The plan shows the estimated credits for every step before you approve, so you know the cost first.
How long can the videos be?
A single clip can run up to 30 seconds. Ask for up to 60 seconds and the agent plans it as several scenes and offers to combine them. For longer projects, send the scenes to the Video Editor.
Can it keep the same character in every scene?
Yes. Upload a character image or let the agent create a character sheet first. Every later scene uses it as a reference, and each scene continues from the previous one.
Can I use my own images, videos or a YouTube link?
Yes. Attach images, clips and audio, pick files from My Stuff, or paste a YouTube link and ask for a new video in the same style.
Where are my results saved?
Generated media is saved to My Stuff automatically. You can also save the whole conversation, rename it and load it again later with its results.
Can I still control every setting myself?
Yes. Turn AI Agent Mode off to use the manual composer and choose the model, aspect ratio, duration and workflow yourself: start and end frame, reference, motion control, edit or upscale.