AMD Developer Hackathon ACT II

AI incident command for modern operations

Inspired by Japan's Andon system, AndonAI detects operational issues, coordinates AI agents, and generates audit-ready response reports.

Live Andon Board5 incidents monitored

Blocked Exit

Zone A-7

Missing PPE

Zone B-3

Overheating

Zone C-12

Defect Pattern

Zone QA-4

Handover Risk

Zone OPS

+ New Alert

Zone ...

From Andon pull-cord to AI-native incident response

In Toyota's production system, any worker can pull the Andon cord to signal a problem and stop the line. AndonAI brings that philosophy into the AI era — where intelligent agents continuously monitor operations, instantly detect issues, and orchestrate the response before problems escalate.

How AndonAI works

Detection

AI-powered visual and sensor monitoring detects operational anomalies in real-time across all zones.

Explanation

Multi-agent AI pipeline interprets issues, matches SOPs, and identifies root causes automatically.

Response

Intelligent severity scoring, team assignment, and escalation routing for immediate incident response.

Reporting

Audit-ready incident reports with evidence, findings, corrective actions, and compliance documentation.

Powered by AMD

Built for AMD Developer Cloud & Instinct MI300X

AndonAI is designed to leverage AMD's open ROCm platform and Instinct MI300X accelerators for high-performance multimodal inference, batch video-frame analysis, embedding generation, and parallel agent workloads — delivering real-time incident detection at scale.

14.1x faster

Multimodal Inference

14.1x faster

Batch Frame Analysis

18.9x faster

Embedding Generation

Simulated benchmarks — real performance measured on AMD Developer Cloud

Ready to see it in action?

Launch the command board or run a demo scenario to see AndonAI's multi-agent incident response in action.