AI Glasses Brasil · Productivity
Concept in validation
Uzz.Ai
From a look in the field to a trackable action.
Hands-free inspection for electrical panels: the technician captures and reports, AI organizes, a human confirms — and the company gets evidence, an owner, a deadline and a history.

The problem
The evidence exists. The process doesn't.
Today, inspectors take a photo and send it on WhatsApp. The image just floats around — no asset, no owner, no deadline, no history.
Today
- Photo lost in the camera roll, a loose message in the chat
- Context gets lost between people and systems
- Open items with no owner, no deadline and no audit trail
Proposal
- Evidence linked to the right asset
- An issue with an owner, deadline and status
- A history the company can audit
Demo
One electrical panel. One sentence. One issue.
- 01The technician observesAsset: QD-03Location: Block A · Ground floor · Hallway
- 02Captures and reports“Circuits four and five unlabeled. Medium priority.”
- 03AI structures itPossible issue · electrical labelingSuggested severity: medium · image: sufficient
- 04The technician confirms“Do you want to log this issue?”Confirm · Edit · Cancel
- 05The company receives itIssue #1048 · Electrical teamDue: 07/31 · Status: open · Evidence: QD-03
No forms. No app switching. No taking your hands off the job.
How it works
Six steps, one record
- 01SessionThe paired phone handles login, permissions and connectivity.
- 02ContextThe technician selects the asset to inspect.
- 03CapturePhoto on command, through the glasses. No continuous recording.
- 04ReportThe voice describes what the eyes picked out.
- 05SuggestionAI combines image, speech and context — and proposes an issue.
- 06RecordOnce confirmed, the issue enters the platform with evidence, an owner and a deadline.
Governance
AI suggests. The technician decides.
No issue is ever created automatically. Technical judgment stays with the technician — AI only organizes what they decided to log.
The company's platform is the structure that supports and drives operations; the glasses are just the capture point.
Scope
What the MVP does — and what it doesn't
Capture is always started by the user. No facial recognition, no continuous ambient recording.
Current stage
Where the idea stands today
- Problem and architecture defined
- Operational flow designed
- Meta DAT POC published — interactive Android + iOS CI Level A
- Ongoing validation in the Developer Center and Mock Device
- Hands-free POC of the inspection journey
- Integration with the operations platform
Technical POC repositoryiOS Level A evidence
What's next
Confirmed observation. Trackable action.
Glasses capture. The phone coordinates. AI structures. A human confirms. The company acts. The hands-free POC is the next chapter — and it's exactly what we want to demonstrate.
