This article explains how to run AI analysis afterward on video materials that have only been uploaded and have not had AI analysis run yet. Use it in situations such as “I want to register the videos for now and run AI analysis on them all together later,” or “I uploaded a video without narration, but I want AI to analyze it after all.”
To learn what each option changes and how to choose one, see AI processing options and how to choose them.
Videos that already had AI analysis run during upload are not covered by this article. If processing failed, see How to re-run a failed AI process. If you want to redo a result that was processed successfully, see “Supplement: When re-running a successfully processed result” in the same article.
Steps to run it
- Open [Materials] in the sidebar
- Click a video that has not been analyzed by AI yet to open the material’s detail modal
-
Click the [Run AI processing] button that appears at the bottom of the detail modal.
- The AI processing options modal opens. Adjust the following as needed:
- Videos without narration (analyzed only by video): Check this box for videos with no narration, videos with only background music or machine sounds, or videos where the speech is unclear
- Recognition Language (in [Advanced settings]): The language the speakers use in the video. Matching it improves the accuracy of the transcription
- Analysis instructions: Entering what kind of video it is improves analysis accuracy
- Split granularity: How finely the steps are split (Process / Step / Action / Single). If you will issue it as an SOP, “Step” is recommended
- Target step count (in [Advanced settings]): Specify roughly how many steps to split the whole video into, or the number of seconds per step. With the default [Auto], the AI decides based on the split granularity
- Reference SOP (in [Advanced settings], Beta): Specify an existing SOP file, and the video can be split to match the sections of that document (for details, see Specifying a reference SOP for video analysis (Beta))
- Text style: The sentence ending of the step text (Imperative / Polite / Suggestive). If you have set up How to write SOPs, you can also select [How to write it] (default)
- Subtitle style: The sentence ending of the action subtitles in the video. It can be set separately from the step text (default is [Imperative]). To match the step text, select [Match text]
- Click [Run], and the status changes to “Waiting” and AI analysis starts in sequence
Analysis usually takes about as long as the video itself. For long videos, analysis continues even if you close the app, and you will be notified by email when it is complete.
Starting from the AI status tag
You can also click the AI status tag in the upper right of a material card (for videos not yet analyzed, it shows “AI not run” or similar) to reach the same detail modal and AI processing options. This is handy when you want to add several videos from the material list to the analysis queue one after another.
Frequently asked questions
Q. Can I analyze several videos at once?
A. You click [Run AI processing] and then [Run] for each video, one at a time. Once run, the videos are queued in order, so you can process them in a batch by simply opening the detail modal of the next video and pressing the button.
Q. How can I adjust the AI processing results later?
A. After analysis is complete, go to [View AI analysis results] → [Re-run AI] in the detail modal. You can change the options and analyze again (the processing overwrites the previous result). For the steps, see the supplementary section of How to re-run a failed AI process.
Q. How can I skip AI analysis when uploading in Step 1?
A. On the AI processing screen during upload, if you cancel without pressing [Run], only the video is registered, without AI analysis. You can analyze it later by following the steps in this article.