In Symphony Creative Studio, Symphony Agent offers two generation modes: Manual Mode and Auto Mode. Auto Mode lets you skip step-by-step confirmations and go directly from your input to finished videos. Instead of reviewing and confirming the product brief, insight report, and storyboard at each stage, Auto Mode handles these steps automatically in the background, delivering your videos faster.
Faster time to video: Go from prompt to finished video in about 5–6 minutes, compared to 15+ minutes in Manual Mode.
Fewer clicks: No need to confirm intermediate steps, the agent handles product analysis, storyboard creation, and generation automatically.
Batch generation: Generate up to 3 video variations in a single session, each with a different creative angle, so you can compare and pick the best one.
Preset video preferences: Set your preferences for voiceover, music, captions, and number of videos upfront, so every generated video matches your requirements from the start.
Full control preserved: You can still refine results by chatting with the agent after videos are delivered, or switch to Manual Mode in a new session for step-by-step control.
See a comparison of Auto Mode vs. Manual Mode:
Feature | Auto Mode | Manual Mode |
Intermediate steps | Brief, insight report, and storyboard are generated in the background (not shown in chat) | Each step (brief, insight report, storyboard) is shown for review and confirmation |
Batch generation | Generate up to 3 videos per session | One video at a time |
Speed | ~5–10 minutes from prompt to video | ~15+ minutes with confirmations |
Best for | Users who want speed, volume, and trust the agent's creative decisions | Users who want detailed control over every creative decision |
Log in to Symphony Creative Studio on your desktop with your TikTok for Business account.
Hover over the left navigation panel and click Agent.
Before starting a new chat, toggle the mode switch from Manual to Auto.
Set your Video Preferences in the prompt box area:
Music: Choose AI-recommended, Yes, or No.
Voiceover: Choose AI-recommended, Yes, or No.
Language: If you chose AI-recommended or Yes for your voiceover setting, select a target language.
Number of videos: Select the desired number of videos in this batch generation.
Enter your prompt. You can:
Describe your product or service in natural language or paste a detailed script.
Upload a product image.
Select a top ad trend, template, or upload a reference video for inspiration.
Click the prompt send button. The agent will automatically:
Analyze your product.
Match to creative strategies or generate a storyboard.
Generate your video(s).
Review your generated video(s). You can then:
Click Regenerate with same storyboard to get new variations from the same storyboard.
Click Generate new variations to get videos with different creative angles.
Pick one specific video to make revisions.
Type feedback in the prompt box to ask for specific changes (for example, "change the hook," "remove the voiceover," "make it more upbeat").
Once satisfied, click Export to download or sync to TikTok Ads Manager.
Note: You cannot switch between Auto Mode and Manual Mode during an active chat session. To change modes, start a new chat.
When you set the number of videos to 2 or 3, the agent generates multiple video variations in parallel. How it creates variations depends on your input:
If you provide a detailed storyboard: The agent generates multiple videos from the same storyboard with different visual seeds, giving you varied executions of the same creative direction.
If you select a reference video or template: The agent creates multiple storyboard variations from the same reference, each emphasizing a different creative element (e.g., different hooks, CTAs, or selling points).
If you provide only a product description: The agent matches your product to multiple creative strategies (potentially inspired by trends, templates, top ads, and your textual description) and generates a unique video for each.
After delivery, you can ask the agent to regenerate all videos, regenerate a specific video, or request changes to any video by referencing it in your prompt.