The brief

Analyst view

• Deploys AI vision systems to catch defects before test stage.
• NVIDIA: 14 days faster builds. Meta: 900+ weeks saved annually.
• Combines line cameras with AI to find known and unknown defects.

Technical approach

As reported

Computer vision

AI-powered visual inspection

Cameras on assembly lines capture images. Machine learning detects known and unknown defects.

Edge inference

Real-time defect detection

Instrumental runs inference on NVIDIA GPUs at the line to intercept defects before test stage, eliminating downstream costs and delays.

Data and analysis

Analyze and improve

Analysis Engine combines factory data across lines. AI agents surface root causes and improvements.

Integration

NVIDIA Metropolis partner

Instrumental joined NVIDIA Metropolis in 2024 to leverage NVIDIA software and hardware for vision AI, accelerating deployments and improving performance.

Show all 4