Why photo review matters
Photo review is the dispatcher's single most important daily workflow. Every time a driver starts or ends a shift, they capture a set of inspection photos in the Fleet by Elevera mobile app. These photos create a visual record of the vehicle's condition. Your job is to compare them against the previous shift and give each shift a verdict. This is how Fleet by Elevera catches new damage early, attributes it to the right shift, and protects your organization from disputed claims.
Clear the review queue every day, ideally first thing in the morning. Unreviewed shifts create gaps in the chain of custody — if damage appears two shifts later and the intermediate shift was never reviewed, it becomes much harder to determine when and how the damage occurred.
Find the shifts waiting for you
- Click Inspections in the sidebar. The badge on the item counts the shifts waiting for a verdict.
- Stay on the Shift reviews tab and the Pending Review list.
- Pending shifts are grouped under Today and Overdue. Shifts the AI flagged come first in each group.
- Click Review on a row to open that shift.
A shift is pending when it has inspection photos and nobody has reviewed it yet. Each row shows the vehicle, the driver, the date, the photos and the AI photo analysis.
Review a shift — step by step
- Compare This shift with the Prior shift for the same vehicle. The page sets each angle side by side, so a new scratch or dent stands out.
- If a frame shows something, click Flag this photo on that angle. The decision bar at the foot of the page counts the flagged frames.
- Read the AI Pre-screening section and the Driver Notes, if there are any.
- In the decision bar, choose a verdict: OK, Suspicious or Damage.
- Add a note in the comment field if you saw something. Be specific: "New scratch on the left rear quarter panel, not in the prior shift" helps far more than "damage found".
- Click Submit Review. The shift leaves the pending list and moves to Review History.
Submit Review stays grey until you choose a verdict. For Damage it also needs at least one flagged frame — Fleet asks you to flag at least one damaged photo before submitting.
The three verdicts
| Verdict | When to use it |
|---|---|
| OK | No damage detected. The vehicle is cleared for the next shift. |
| Suspicious | Possible damage. Flag it for follow-up without escalating yet — for example, arrange a physical check before the next shift. |
| Damage | Confirmed damage. Flag the frames that show it; submitting opens the recovery work described below. |
What a Damage verdict starts
Submitting a Damage verdict does two things in the same step:
- Fleet opens a recovery case for the damage in the Recovery workspace, owned by you, with a 24-hour deadline.
- The Report incident wizard opens with your flagged frames already attached, so you can record the incident right away. Completing it links the incident to the recovery case.
How to follow the case from there is covered in From a Damage verdict to a recovery case.
Review faster with the queue
The Start review queue panel at the top of the page walks you through every pending shift, one after the other. Click Launch queue to begin. Today's shifts come first, then overdue ones; within each, the shifts the AI flagged lead.
Keyboard shortcuts in the queue:
| Key | Action |
|---|---|
1 | Choose OK |
2 | Choose Suspicious |
3 | Choose Damage |
Enter | Submit the chosen verdict |
← | Skip this shift |
Esc | Leave the queue |
Shortcuts pause while you type in the comment field.
AI-assisted vehicle condition assessment
When AI photo analysis is enabled, Fleet examines the inspection photos and provides advisory context in the AI Pre-screening section:
- Review bucket — Clear, Inconclusive / retake, or Possible damage
- Confidence score — A percentage (0–100%) describing the model's confidence in that bucket; it is not a damage probability
- Reasoning — A short explanation of what the AI observed or why the evidence is inconclusive
- Photo-level flags — The affected photo, damage type, severity and vehicle area; Fleet does not draw a bounding box
If a shift has not been analysed yet, the section says so and you review the photos on their own.
The AI analysis is an assistant, not a damage diagnosis. Always review the complete photos yourself. The AI may miss subtle damage and may flag reflections, shadows or dirt as possible issues.
Every shift needs a human decision
AI does not close inspections. A Clear result stays in the pending list until a dispatcher checks the evidence and submits a verdict. Historical review entries may still show an auto-reviewed marker from an earlier workflow.
Review history
The Review History tab lists every submitted review with the shift date, vehicle, driver, verdict, comment, who reviewed it and when. It is your audit trail: if a question comes up about when damage was first noticed, trace it back here.
To correct a verdict, use the verdict buttons on that row. Changing a verdict to Damage opens a recovery case and the incident wizard, just as a new Damage verdict does.
Comparing with previous shift photos
The power of the review system comes from the chain of comparisons. Each review compares against the most recent previous shift that has photos. Over time, this creates a continuous visual record:
- Shift A photos (baseline) → reviewed as OK
- Shift B photos compared against Shift A → reviewed as OK
- Shift C photos compared against Shift B → reviewed as Damage (new scratch)
This chain makes it clear that the scratch appeared during Shift C, which determines which driver was responsible for the vehicle at the time the damage occurred. If you break the chain by skipping reviews, you lose this ability. That is why clearing the queue daily matters.
If no earlier shift with photos exists for the vehicle, the prior side stays empty. Review the current photos on their own and document any damage that is already there.
Tips for efficient reviewing
- Review in chronological order when catching up on a backlog, so each comparison is meaningful.
- Use the AI bucket as a triage aid, not a shortcut. Inconclusive and Possible damage results deserve careful attention; Clear results still need a human check.
- Flag specific frames rather than leaving a bare verdict. They travel with the review into the recovery case and the incident.
- Pay special attention to corners, wheel arches, bumpers and mirrors — the most common places for new damage.