This release focuses on AI check (double verification, beta) for inspection and confirmation questions, along with new features and improvements. AI re-examines photos submitted by workers, making it easier to spot differences from the model and potential issues. We've also added 3D editing (beta) to position AR content without holding a device, step merging, text hiding, and other feature updates.

① Added AI check (double verification of submitted photos) — beta

For questions requiring photo submission, we added AI check — a second pass where AI re-examines photos workers submit. This adds a second set of eyes powered by AI to human verification.

This feature is provided as a beta. Specifications and judgment accuracy will improve over time. If you notice examples that worked well or cases where judgment didn't match intent, please send feedback to customer-success@episotech.com (Dive Customer Success).

●How verification works (2 methods)

Compare with Model image: When you register a Model image for a question, AI compares the submitted photo against it
Verify using question and step content: Without a Model image, AI checks the photo based on the question text and step descriptions

●Result types

Match confirmed / No issues found (advisory opinion only — final confirmation is your responsibility)
May differ from Model — please verify again (check once more)
Cannot determine from photo (retake photo with target clearly visible, then compare)

Results are advisory opinion only. They do not replace final decisions like "Passed" or "Compliant" — use them as a supporting tool with verification by the worker or manager.

●Setup (SOP creators)

For questions with required photo submission, enable 「AI check」 and register a Model image as needed. The Model image displays below the question text, and for pass/fail judgment questions, workers also see the "Normal" reference standard. See How to use AI check (double verification of submitted photos, beta) for detailed setup and usage.

●Notes

Available to Pro and Enterprise Plan customers
Works only on questions requiring photo submission (check is available after photo upload completes)
AI check unavailable Offline (resumes when back Online)
Does not display on unconfigured questions; existing questions behave as before

●AI check (comparing with Model)

② AR content can now be edited without a device and viewed in 3D even on unsupported devices — 3D editing, beta

Previously, positioning AR content (arrows, text, images, audio, 3D models, etc.) centered on 「Edit AR」 using the mobile app with device in hand. We've now added 「3D editing (beta)」— positioning and editing AR content in 3D space from your browser on PC or smartphone, without holding a device. You can proceed with placement work from your desk even without the physical item on hand.

We've also added 「View in 3D」 to the viewing screen. Even on devices that don't support AR (holding up a device), you can view AR content in 3D, rotating and zooming to inspect it.

When you turn 「Use」 ON in the AR Settings for the target step, 「3D editing (beta)」 button appears next to 「Edit AR」
Creating and editing AR content (Edit AR / 3D editing) is available for Pro and Enterprise Plan customers
「View in 3D」 (viewing) is available to anyone who can view that SOP

「3D editing」is provided as beta. For detailed usage, see 3D editing (position AR content in 3D space). For where AR fits best, also review Where AR works best and how to embed it in steps.

●View in 3D (viewing screen)

③ SOPs can now show audio and images only — hide text

We've added a setting per SOP to not display description text for each step on screen during playback. Use this when you want images or slides shown with narration audio only, like a video.

Set this in the SOP edit screen under 「Playback & display settings」 by turning ON 「Do not display text on screen」 (configured by SOP creators)
When viewing an SOP with this ON, description text doesn't display — instead audio reads it aloud (displays "This procedure is set to hide text")
Viewers can toggle 「Show text」 on the playback screen to temporarily display text on the spot
Description text itself is not deleted (just hidden). Turning the setting OFF restores normal display

For detailed usage, see Hide text (descriptions) during playback.

●Playback & display settings > Do not display text on screen

④ Step merging now preserves descriptions

When merging adjacent steps, the description text from the later step is now automatically carried over (transferred) to the first step. Merging doesn't lose the explanation — it reads as one complete step.

Unlinking reverses the carry-over, returning it to the original step (no duplication)
If text was manually edited after carry-over, we don't overwrite it (safer behavior)

⑤ Other improvements and bug fixes

●AR and 3D content saving

Fixed a bug where placed images, audio, and 3D models fail to display (don't load) after saving
Fixed a bug where placed content (3D data) disappears when re-editing and saving created AR/3D content
Also fixed other bugs around Media saving in AR/3D content

●Other

Changed "Voiceover" to "Audio" (feature is the same; display name only)
Fixed a bug where approval workflow targets couldn't approve when an SOP was shared (Draft shared)
Fixed a bug where pressing Enter in a text field loses focus and prevents line breaks
Fixed a bug where Folders could be mistakenly specified at the top level (directly under Root)

When settings are unchanged, existing SOPs and analysis results behave as before. AI check displays only when enabled for the target question. "Voiceover" changes to "Audio" in the display name, but the feature is the same.

Related information (how to use each feature)

◎For any questions about this update, feel free to Contact us at customer-success@episotech.com.