We open-sourced our Rust IoT stack because "trust us" doesn't work in healthcare (www.teton.ai)

Teton announced it has open-sourced Smith, a Rust-based fleet management platform it built to deploy, monitor, and update thousands of AI-powered computer-vision sensors across hospitals, care homes, and senior living facilities. Smith—released quietly as v0.2 last month—was born from a 10x scale-up (hundreds to thousands of devices) and enforces safety-critical guarantees such as >99% uptime and seamless upgrade/rollback. The team open-sourced it because transparency builds trust with hospital IT and clinicians: in healthcare “trust us” isn’t enough when device downtime can directly affect patient safety. For the AI/ML community, Smith is significant because it tackles the often-overlooked edge-operational problems of deploying and maintaining ML-driven sensors at scale. Key technical points: Smith is implemented in Rust for performance/safety, focuses on deployment orchestration, monitoring, and resilient updates, and is being developed toward a stateless, extensible model with a CLI and dashboard akin to Kubernetes tooling. Open-sourcing invites contributions on reliability, security, and automation—helping teams running edge CV models reduce manual ops, improve compliance, and share best practices for safe, large-scale model rollout in critical environments.