Wearable
Camera and audio provide first-person field context.
USC FIELD TECHNOLOGY / CONTROLLED PILOT
How We’re Testing AI-Assisted Field Safety in Sacramento
US Construction & Remodeling Corp. is testing a human-reviewed workflow that connects Meta smart glasses, an iPhone-based USC Assistant, and our internal business systems. The goal is simple: help authorized field teams capture context, ask for help, and escalate potential concerns more clearly.
PILOT STATUS · JULY 2026
Connected field tools. Clearer communication. Human-led decisions.
Pilot brief / Sacramento / July 2026
A field report, not another service-page template.
This controlled pilot explores how wearable capture can help an authorized worker explain a field question with better visual context and reach the right reviewer faster.
“Capture what matters. Route it to the right person. Keep the decision human.”
- Device layer
- Meta smart glasses + iPhone
- Business layer
- USC Assistant + ERP
- Current state
- Controlled pilot
- Decision owner
- Qualified people
OSHA treats hazard identification as a proactive, ongoing process and recommends documenting problem areas with photos or video for discussion and learning. OSHA hazard-identification guidance · OSHA training tool
Related USC workflow: Our Remote Roof Assessment uses guided photo review to help homeowners organize visible roof conditions before on-site confirmation.
The signal path
Five handoffs.
One accountable workflow.
The glasses provide the field view. USC’s mobile and ERP systems connect each question to authorized people, project context, and a traceable business process.
USC iPhone app
The mobile bridge manages connection, permission, capture, and audio.
USC Assistant
The question and media become one organized conversation.
USC ERP
Identity, authorization, workflow, and records stay in the business system.
Human review
A manager or qualified person evaluates the context and chooses the next step.
Evidence ledger
What we can prove today.
The page separates verified tests, active pilot work, and third-party dependencies. That is more useful than a vague “AI-powered” claim.
Verified
Real iPhone app
A signed USC iOS app launches on a physical iPhone, and the assistant interface displays both user and assistant messages.
Verified
Glasses connection
The physical glasses were detected as connected and compatible. Basic audio routing and push-to-talk behavior were tested.
Verified
Prototype media path
Mock-device tests verified capture, conversation attachment, and synthetic audio return. A prior controlled developer-mode test routed a real glasses photo into USC conversation.
Pilot
Field reliability
Reconnect behavior, error recovery, worker training, privacy controls, and limited jobsite use still need repeatable validation.
In progress
Meta provisioning
Repeatable physical photo capture is being validated as Meta’s developer-preview provisioning continues to evolve.
Field sequence
From “look at this” to a traceable next step.
A worker has a real question
The flow starts because an authorized person needs help or wants to document a specific condition.
Only the needed context is captured
The user avoids unrelated people, private areas, addresses, screens, and confidential documents whenever possible.
Media joins the USC conversation
The iPhone app attaches the visual context to the assistant workflow instead of leaving it as an isolated photo.
AI organizes—people evaluate
The assistant summarizes details or flags a potential concern so a qualified team member can evaluate it and choose the next step.
The response becomes a record
The question, media, human response, and next step can remain associated with the relevant business process.
The human boundary
Assistance on one side.
Authority on the other.
AI can assist with
- Organizing a field question and its visual context
- Summarizing visible details for review
- Flagging a potential concern that deserves attention
- Making the exchange easier to trace
People remain responsible for
- Evaluating the actual jobsite condition
- Stopping, continuing, or redirecting work
- Selecting controls and applying company procedures
- Code, safety, engineering, and trade decisions
PurposeCapture only for a defined work question.
AccessLimit the workflow to authorized users.
MinimizeAvoid unnecessary people and private information.
ReviewDocument retention, deletion, and vendor handling before expansion.
The workflow follows practical privacy principles: clear purpose, limited access, minimal collection, and responsible review. California Privacy Protection Agency FAQ.
AI-assisted field safety
Smarter support for the people in the field.
Smart glasses and the VS Assistant help field teams capture visible concerns, share useful context, and move faster from observation to review.
Named author / traceable method
The person and method behind the page.
Pilot FAQ
Common questions about the workflow.
Is this system already deployed on every US Construction jobsite?
No. This is a controlled pilot, not a company-wide production rollout. We have verified the signed iPhone app, assistant interface, compatible glasses connection, basic audio route, and prototype media workflow. Wider deployment depends on repeatable field testing and the stability of Meta's developer-preview provisioning.
Does AI decide whether a construction condition is safe?
No. The assistant can organize visual context and flag a potential concern for human review. A qualified person remains responsible for evaluating the site, choosing controls, stopping work when necessary, and following the applicable safety program and regulations.
What can a worker do during the pilot?
The intended workflow lets an authorized worker capture field context, ask a question, and route the photo and conversation to the USC Assistant. A manager or qualified reviewer can then examine the information and respond through the existing company workflow.
Has photo transfer from the glasses been tested?
Yes, in a previous controlled developer-mode test a real glasses photo reached the USC conversation. Mock-device tests also verified photo capture, conversation attachment, and synthetic audio return. Current physical capture is not presented as a generally available feature because Meta provisioning must first become repeatable.
How does USC handle privacy in a camera-based field workflow?
The pilot is being designed around authorized users, a defined work purpose, limited access, human review, and avoiding unnecessary capture. Images can contain personal information, so notice, access, retention, deletion, and vendor handling must be reviewed before broader deployment.
Is Meta sponsoring or endorsing this pilot?
No sponsorship, partnership, or endorsement is implied. Meta and its product names are third-party trademarks. USC is independently evaluating Meta smart glasses and developer tools as one device layer in its own controlled workflow.
What happens next?
The next phase is to revalidate physical photo capture, document repeatable setup, test limited field scenarios, review privacy and safety controls, and measure whether the workflow improves escalation quality without distracting workers or replacing established procedures.
Pilot conversations / Sacramento
Bring us a real field problem.
We welcome practical conversations with construction, safety, technology, education, and workforce organizations across the Sacramento region.
Privacy Policy and learn how USC handles project information.