AI Hazard Marking Tutorial
This walks through one photo, from the moment you open it to the corrective actions it leaves behind. See AI Hazard Marking for what each part does.
Before you start
- Your company needs the image analysis feature and AI budget left for the month.
- You need the same permission that lets you annotate a photo, because marking replaces the stored photo.
- To create actions from the hazards, you also need the Actions module, permission to add findings, and an active project.
Mark the hazards on a photo
- Open the record the photo belongs to — a finding, an incident or an inspection — and tap the photo to open it full size.
- Tap the AI button to open the analysis sheet.
- Switch on Mark hazards on the photo. The row below the switch tells you it uses more AI budget and replaces the original photo.
- Optionally type what you want the AI to look at. Left empty, it describes the picture and gives safety recommendations.
- Tap Inspect and wait. The answer comes back with the analysis, a list of the hazards, and the marked photo.
You now have a photo with a numbered box around each hazard, coloured by its risk level on your company's matrix, and a key at the top of the photo's short description explaining what each number is.
Tip
Check the boxes before you move on. If one sits in the wrong place, tap Restore original photo — the original comes back and the key is removed from the description. You can run the analysis again afterwards.
Turn the hazards into actions
- Under the result, find the Turn the hazards into actions card and tap Review actions.
- The sheet lists one action per hazard. For each row, check:
- the task — the fix the AI suggests; edit the wording if you would put it differently
- the priority and the due date — set from the hazard's risk level, and changeable per row
- the type of control — where the fix sits in the hierarchy of controls
- Untick any hazard you do not want an action for.
- Tap Create actions. The count on the button is how many will be created.
Each action is created with the marked photo attached and linked to the incident or inspection the photo belongs to, so it shows up where the work is tracked. The card then reads Actions created, so the same hazards are not turned into actions twice by accident.
The quick route: HSE Support
When there is no record yet — you are standing in front of the problem with your phone — use the chat instead:
- Open HSE Support and tap the attachment button, then Take photo or choose one from the gallery.
- Switch on Mark hazards on the photo, shown under the photo preview.
- Tap send. You do not have to type anything.
- The assistant replies with the marked photo and the numbered list of hazards. Tap the photo to see it full size.
- Turn the hazards into actions appears under the reply and opens the same review sheet.
Actions created from a chat photo carry the marked photo, but belong to no incident or inspection — a chat photo belongs to neither. Your original photo stays in the conversation, so there is nothing to restore.
Reading the result
- Red, amber, green follow your company's own risk matrix, not a fixed scale.
- The photo carries the numbers; the names and ratings live in the photo's description, and travel with it into reports.
- At most ten hazards are marked on one photo, ordered from highest to lowest risk.
Warning
The AI can miss a hazard or misplace a box, and a drafted action is a starting point, not a decision. Read both before you keep them.
What's Next?
- Photo Documentation tutorial: attaching, ordering and annotating photos
- AI Hazard Marking: the full feature reference
- HSE Management tutorial: where the actions you created are tracked