When you create an SOP from a video or file, you can specify how the AI should analyze it in “AI processing options”. Even if you press [Run] with the default settings, the SOP will still be created. This page explains what each item changes and how to choose it.
AI processing options are available on all plans. The number of AI analysis runs depends on your plan.
Open AI processing options
The same screen opens in any of the following ways.
- It appears automatically when you upload a video or file
- For materials that have not been analyzed by AI, open it from [Run AI processing] in the material details window (Run AI analysis on video material that has not been analyzed)
- To redo an already analyzed material, open it from [Check AI analysis results] and then [Re-run AI] (How to re-run failed AI processing)
The screen is divided into three sections: [Analysis instructions], [Analysis options], and [Advanced settings]. You usually adjust the first two. The question settings (question type), the target step count, and other options are under [Advanced settings].
Analysis instructions
This is where you tell the AI what you want it to know about this video. Writing down the target equipment, the purpose of the work, the tools used, and so on, will improve the accuracy of the analysis.
In addition to what the work is, you can also tell the AI which scenes to keep, how to write the text, and how to create questions (Advanced analysis instructions: keep NG cases and retries for training materials (customer service role-play example)). For options you can choose in [Analysis options] or [Advanced settings], such as the target step count, text style, and question type, please select them there.
You only need to write what is specific to this video. Common premises that apply to every video, such as in-house technical terms and equipment names, only need to be registered once in the Terminology Glossary, and they will be reflected in AI analysis without being entered each time.
You can also insert a saved template from the icon at the top right of the field.
Analysis options
Split granularity
Specify how finely the video is divided into steps.
| Option | Unit of division |
|---|---|
| Process | Divides into large units such as preparation, execution, and cleanup |
| Step | Divides into standard work-procedure units. This is recommended when publishing as an SOP |
| Action | Divides finely into individual actions, such as button operations and tool use |
| Single | Does not divide. Treats the whole video as one step |
The default is [Step]. Choose [Single] if you want to use the video as one edited video without dividing it into steps (How to create one edited video without splitting it into steps).
Text style
Specify the sentence endings and wording of the step text.
| Option | Description |
|---|---|
| Imperative | Plain imperative form. This is the standard for SOPs |
| Polite | Polite form. For new-employee training, or for SOPs submitted to business partners or auditors |
| Inviting | Soft, inviting tone. For education and training |
The default is [Imperative]. If you have registered How to write SOPs, the option [How to write SOPs] is added and becomes the default.
Subtitle style
Specify the sentence endings and wording of the action subtitles shown on the video. This can be set separately from the step text. The options are the same three as the text style, plus [Match text]. The default is [Imperative].
If you select [Match text], the subtitles follow the text style of the step text. However, when the text style is [How to write SOPs], [Match text] cannot be selected, so choose the subtitle style separately.
Videos without narration (analyzed only by video)
When checked, the AI ignores the audio and analyzes only the video. Step division and action descriptions are generated as before, but no speech transcript is generated.
Use this for videos without narration, videos with only background music or machine sounds, or videos where the speech is unclear. It is also useful when you re-run a video whose AI analysis did not work well because of the audio.
If a video has no narration explaining the work and you analyze it without this setting, the step splits may be divided mechanically into equal time segments instead of at the breaks in the work. The AI uses changes in the topic of the narration as one of its clues for where to split. If the speech is unrelated to the work, there are no clues, so the AI ends up aiming only at the target number of steps.
If the analysis results are in this state, a notice saying “The step splits look mechanical” appears just before you create the SOP.
- Creation does not stop. For a steady, fixed-pace workflow, the equal division may actually match reality, so you can choose [Create as is]
- Pressing [Re-analyze] opens the AI analysis options, with [Videos without narration (analyzed only by video)] shown in a visible position. Please decide for yourself whether to check it
- Re-analysis started from this notice is not counted toward your AI processing runs
Advanced settings
Create questions (when creating an SOP from a file)
This is shown when you create an SOP from a file (PDF or Office file). When checked, the AI creates questions for the worker from the content of the material. It is ON by default.
When OFF, only the questions written in the original material are imported (the previous behavior). No questions are created for pages without work content, such as the cover, table of contents, or revision history. Questions you have created can also be removed with [Apply questions to steps] under [Step conversion settings] when you create the SOP.
When [Body writing style] is set to [Transcribe], this option cannot be selected, because that mode copies the original text as it is. In this case, only the questions written in the original material are imported.
Question type
Specify how the AI creates questions. This is always shown when creating an SOP from a video. When creating an SOP from a file, it is shown only when [Create questions] is checked.
| Option | Description |
| On-site check | Creates questions that let the worker check their own work on the spot. Only choices that could actually happen in that process are listed, and no false choices that do not happen are created. It can be used as a digital version of pointing and calling (shisa-kanko) |
| Comprehension test | To test memory, false choices that do not actually happen are also mixed in. Choose this when used for education or to check proficiency |
The default is [On-site check]. The SOP player does not give scores or pass/fail results, and it is built so that the user “cannot continue until they answer.” So, to avoid the user getting stuck halfway through the steps because they cannot remember, this is the default setting.
The option chosen here also carries over to the question settings when you create an SOP. For materials analyzed with [On-site check], [Continue when unexpected] under [Step conversion settings] is turned ON from the start.
Target step count
Specify a guide for the number of steps. The default is [Auto], and the AI decides the level of detail based on the split granularity.
| Option | Description |
|---|---|
| Auto | The AI decides based on the split granularity (default) |
| Specify step count | Specify roughly how many steps the whole video should be divided into |
| Specify seconds | Specify roughly how many seconds each step should take |
Specifying these does not change how steps are divided (at meaningful breaks in the work). The target only affects the number of steps. Also, it is a target, and it does not guarantee that the number will be exact.
When the split granularity is [Single], this is not shown, because it conflicts with the setting of one step for the whole video. When you match the breaks of the material with [Reference SOP], the number of steps in the material takes priority, so this setting is not used.
Recognition language
This is the language spoken in the video. Setting it correctly improves the accuracy of the transcript. Languages not on the list can also be searched for by typing them directly into the input field.
Aggressively cut sections without speech
This setting removes a wide range of sections without speech (narration) from the SOP being created. The default is OFF.
Even when OFF, obviously unnecessary parts such as filler words and long silences are cut conservatively, but natural short pauses during the work are kept. If you turn it ON for a video where the work continues in silence, the work itself may be cut, so the default is OFF.
Turn it ON when you want to tighten up slow parts in videos you recorded while explaining, or in videos with continuous speech such as lectures or presentations. Cut sections can be restored in the video editor.
If [Videos without narration (analyzed only by video)] is checked, this setting cannot be selected, because it assumes there is no speech.
Reference SOP (beta)
You can specify one existing SOP file (PDF, Excel, Word, or image), and the video will be divided to match that file's breaks, number of steps, and order. Use this when you want to carry over the structure and terms of an SOP you have already prepared into a video manual.
When you specify a reference SOP, the split granularity and the target step count are not used. For details, see Specify a reference SOP for video analysis (beta).
How to choose between split granularity and target step count
| What you want to do | Setting |
|---|---|
| Let the AI make the SOP | Split granularity = Step, Target step count = Auto |
| Split more finely or more coarsely | Change split granularity to Action or Process |
| Make the number of steps even | Specify the step count or seconds with Target step count |
| Match the breaks of an existing SOP | Specify a reference SOP |
| Do not want to split into steps | Split granularity = Single |
Creating an SOP from a file
When creating an SOP from a file (PDF, Excel, Word, or PowerPoint), the items are different from those for video. Split granularity, subtitle style, target step count, and reference SOP are not shown.
| Item | Description |
|---|---|
| Conversion method | These are the page settings for Office files. Choose from [1 sheet = 1 page], [Fit width to 1 page], or [Print settings]. If the print settings of the original file are already set up properly, [Print settings] is recommended |
| Step split method | Sets how many steps are created from one page. [Per content] splits table rows and bullet points within a page into steps. [Per page] makes one page into one step. In both cases, each step fits within one page and does not span pages |
| Body writing style | [Organize] reads the content and organizes it into work content, tips, cautions, and safety. [Transcribe] copies the original text as it is. Use [Transcribe] for materials that put the original wording first, such as regulations or check sheets |
| Create questions | The AI creates questions for the worker from the content of the material. The default is ON. When OFF, only the questions written in the original material are imported (the previous behavior). This cannot be selected when [Body writing style] is [Transcribe] |
| Question type | Shown only when [Create questions] is checked. The options are the same as for video (On-site check / Comprehension test) |
For details, see Create an SOP from a file (Excel, Word, and PowerPoint supported).
Remembered settings
Your selections are remembered and become the initial values the next time you open this option screen. When you [Re-run AI] on an analyzed material, the conditions used the last time it was analyzed (such as the reference SOP and target step count) are restored.
You can check the conditions under which an analyzed material was analyzed from [Check AI analysis results] for that material.
Frequently asked questions
Q. How long does analysis take?
A. Roughly the same as the length of the video (a 10-minute video takes about 10 minutes). Analysis continues even if you close the app, and you will be notified by email when it is complete. Each video can be up to 1 hour long.
Q. Is there a setting that does not generate subtitles?
A. No. Subtitles are always generated. If you do not want speech transcripts generated, check [Videos without narration (analyzed only by video)].
Q. Can I change the options and analyze again?
A. Yes. From [Check AI analysis results], you can change the options and analyze again with [Re-run AI]. The previous analysis results will be overwritten.
Q. The step splits are not what I expected.
A. First, change the split granularity and analyze again. If you want to align the number of steps, specify the target step count. If you want to match the breaks of an existing SOP, specify a reference SOP. If it still does not match, you can manually edit the steps in the video editor.
Related information
- Step 1: Upload a video
- Run AI analysis on video material that has not been analyzed (apply AI processing later)
- Specify a reference SOP for video analysis (beta)
- Define how to write SOPs (align the organization's wording and standards with the AI)
- Terminology Glossary (terms and multilingual translations)
- Unify spelling variations (Aliases in the Terminology Glossary)
- How to re-run failed AI processing