All AI/ML services with health, latency, error rate, deployments
// TODO: implement using vercel-tokens