wendy.dev · docs.wendy.dev
contact@wendy.dev
Enterprise overview
Building AI for the physical world is unreasonably hard. Until Wendy.
It is not the models. The setup underneath them is brittle: hand-flashed images, SSH scripts, snowflake devices, and updates that brick hardware in the field.
Wendy replaces that foundation with the open-source OS and toolchain for robots, drones, smart cameras, and the sensors and actuators that run real-world operations. Plug in over USB-C, deploy a containerized app with one command, and update devices in the field without fear.
Hundreds of teams deploy with Wendy every day, from startups to enterprises, shipping to production with almost 90% fewer resources.
Quality inspection, predictive maintenance, and site safety across factories and energy infrastructure. Smart cameras, gates, and actuators run offline, and process data never leaves the plant.
Perception, autonomy, and control for contested, disconnected environments. Everything runs on the device: offline, secure, and field-ready, with no cloud dependency.
Clinical-grade AI for regulated hardware. Imaging, monitoring, and inference run directly on the device, private and deterministic at the point of care.
Labs and R&D teams iterate on real hardware in minutes, from bench prototype to field trial, without rebuilding the deployment stack each time.
Run AI on infrastructure you own. Legal, medical, defense, and financial firms deploy models to their own on-premise servers with one command, and data, prompts, and inference never leave the building.
The operating system
A hardened Linux for NVIDIA Jetson and Raspberry Pi. Headless and lightweight, with containerized apps, an immutable system image, and atomic A/B updates that roll back automatically if a boot fails.
The developer tool
A single command, wendy run, takes any app with a Dockerfile from laptop to device over USB-C or the LAN, streaming logs back live. Swift, Python, Rust, C++, and TypeScript, with ROS 2 support.
Fleet management · early preview
Manage, monitor, and update devices from anywhere as you scale from one prototype to a fleet: OTA update workflows, crash reporting, and OpenTelemetry-based remote telemetry.
Apps run as unprivileged containers. Entitlements gate access to GPU, camera, audio, network, USB, GPIO, and storage. Every app gets only what it declares.
Every connection to a provisioned device requires mutual TLS with per-device certificates, with post-quantum cryptography for data in transit. Production images ship without SSH.
OS releases install to an inactive A/B partition. If the device fails to boot, it rolls back to the previous version on its own.
The full stack is Apache 2.0 open source, with a published STRIDE threat model that names threats, trust boundaries, and mitigations.
Runs on: NVIDIA Jetson Orin Nano · AGX Orin · AGX Thor · Raspberry Pi 3/4/5 · any Linux machine via Wendy Agent