Walk-and-talk capture and triage

Walk-and-talk capture is for the stretches where stopping to find the right item slows you down. You capture first and assign each photo or recording to an item later, at the truck or back at the office.

Capture with nothing selected

1. Open the inspection. At the top, above the items, is the Walk-and-talk capture panel with the hint "Dictate or snap photos as you go - sort them onto items afterward."

   

2. Tap Add photo to take a photo or video, or choose existing ones. You can select several at once. Or tap Dictate, speak, and tap Stop recording.

3. Repeat as you walk. Each capture is stored on your device and uploads in the background, so it works with no signal. See working offline.

Nothing is turned into a finding yet. The panel is hidden once an inspection is completed, because a completed inspection is read-only.

Sort the captures

When there is at least one unsorted capture, a Sort captures (N) section appears above the items, with the note "Captured during your walkthrough with no item selected - assign each to an item below."

1. For each row, open the Choose an item... menu. Items are grouped under their section names (Roof, Exterior and so on). Pick the item the capture belongs to.

2. Tap Assign. The row leaves the list.

   

3. Tap Discard instead to delete a capture you do not want.

What assigning does depends on the capture:

• A photo or video goes onto the item's own Photos area. It is not tied to a finding; to put a photo on a finding, add it under that finding. See photos and video.
• A recording becomes a new finding on that item with a severity of Minor. If the recording has been transcribed, its text is the finding's narrative. If not, the finding says "(awaiting dictation)" until the text arrives. Open it afterwards to set the severity and fix the wording. See dictate a finding.

Let the app suggest an item

Two optional buttons use AI to suggest where a capture belongs. They only suggest; nothing is assigned until you tap Assign.

• Classify N photo(s) with AI appears above the list once photos have finished uploading. It looks at all the unsorted photos together, using a small copy of each and the names of the items in this inspection. For each photo it pre-selects an item in the menu and shows "AI suggestion: " followed by the item name and a short reason. If it cannot tell, the row says "AI couldn't tell - pick manually." If you have already picked an item for a photo, the suggestion does not replace your choice.
• AI suggest item appears on a recording row once its transcript is ready. It works from the words in the transcript, and it does select the item it suggests in the menu, so check the menu before you tap Assign.

Until photos have uploaded, the list says how many are "still uploading - AI classification will be available shortly."

If something goes wrong

• "Could not access the microphone." Allow microphone access for the site, then tap Dictate again.
• "Video is 75s, longer than the 60s limit." Videos over 60 seconds are not accepted. Shorten the clip and add it again.
• "Could not process that file." The file could not be read as a photo or video. Try another.
• "Photos are still syncing - wait a moment and try again." Classification needs the photos to have reached the server. Wait for the header to show "All synced" and try again.
• "Could not classify photos. Try again." or "Could not get an AI suggestion." Pick the item yourself from the menu.
• "Transcription failed - assign manually." The recording is kept. Choose an item and assign it, then write the finding by hand.
• "Could not assign this capture. Try again." Tap Assign again.