🛡️

Newmetrix

AI-powered jobsite safety monitoring and hazard detection

Computer vision platform that automatically analyzes jobsite photos and videos to detect safety hazards, track compliance, and predict incidents. Part of Oracle Construction Intelligence Cloud.

$22.4M raised 🏢 Established

Founded 2015 • Major contractors including JE Dunn, Shawmut Design and Construction, Warfel, Barton Malow, Suffolk, Obayashi, Lithko

Website → ← All Tools

Why This Tool Exists

⚠️

The Problem

Construction teams manually review hundreds of daily jobsite photos to identify safety hazards, a process that takes hours and often misses issues. Safety incidents cause project delays, increased insurance costs, and worker injuries—often costing contractors hundreds of thousands per incident.

✅

The Solution

Newmetrix automatically analyzes jobsite photos and videos with computer vision trained on 12 centuries of incident data and 17 million construction images. Its AI engine (Vinnie) detects 100+ safety hazards and predicts which projects are at highest risk within minutes, enabling preventive action.

How You Use It

🚀

Delivery Method

SaaS Mobile App

🔌

Integrations

Procore Autodesk BIM360 Oracle Aconex OxBlue Box Egnyte SharePoint Online

Disciplines

Construction, Safety Management

Project Phases

Construction Administration

Project Types

Commercial, Infrastructure, Mixed-Use, Industrial, Hospitality

Data Transparency

Exactly what this tool uses and how

📥

Input

What it needs

Required: Jobsite photos or videos, Project information

Optional: Procore project data, BIM 360 documents, Project schedule

Formats: JPG, PNG, MP4, Video streams, Mobile photos

📤

Output

What you get

Format: Safety risk assessments and analytics dashboards

Fields: Detected hazards with severity ratings, Safety compliance scores, Risk predictions by project, Incident trend analysis, Worker safety benchmarking, Corrective action recommendations

⚙️

Algorithm

How it works

Model: Proprietary computer vision AI trained on construction safety incidents

Accuracy: Trained on 12 centuries of incident data and 17 million construction images; can detect 100+ safety hazard types (specific accuracy metrics not publicly specified)

🔒

Privacy

How your data is protected

Retention: Photos and videos retained according to project lifecycle (specific retention period not publicly specified)

Training: Not publicly specified whether photos are used for AI model retraining

Compliance: SOC 2 (inferred from Oracle enterprise standards), GDPR (likely, as part of Oracle portfolio)

🔌

API

Integration

Endpoint: https://api.smartvid.io/api/v1/doc (documented in help center)

Method: REST API (POST for authentication, GET/POST for integration operations)

Use Cases

• Automated daily jobsite safety inspections across multiple projects
• Identify and correct PPE compliance issues before incidents occur
• Detect workers at height, scaffolding, and fall risks automatically
• Benchmark safety performance across project portfolio
• Predict high-risk projects and allocate safety resources proactively
• Create objective safety records for insurance and liability protection

Pricing

Free

Not publicly available

Pro

Contact vendor for pricing (likely project-based or per-photo analysis)

Enterprise

Custom pricing as part of Oracle Construction Intelligence Cloud

Website →

📚 Research Sources & Data Quality Last verified: 2026-02-01

Data Sources (10)

✅

Verified Data (21)

vendor, website, yearFounded, maturityStage, customerBase, fundingTotal, problem, solution, deliveryMethod, integrations, disciplines, projectPhases, projectTypes, input-required, input-optional, input-formats, output-format, output-fields, algorithm-approach, control-features, use-cases

❌

Not Found (7)

pricing-details, detailed-privacy-policy, data-retention-specific-duration, soc2-certification-report, model-accuracy-quantified-metrics, api-rate-limits, data-training-practices

Our Commitment: We only include verified data from official sources. If information isn't publicly available, we mark it as "Not publicly specified" rather than guessing.