platoseedplatoseed
The default way of running on-device AI at Scale
Edge AI is inevitable, but shipping it is painful: every device class behaves differently, runtimes vary, models are huge, and performance collapses under memory/power constraints. RunAnywhere turns that into an enterprise-ready workflow: one SDK to run models on-device, plus a control plane to manage models, enforce policies, and measure outcomes across thousands of devices.
RunAnywhere provides on-device AI inference tooling for mobile and edge devices, enabling private, hardware-native AI by running models directly on devices like Apple Silicon and mobile hardware. It offers engines, SDKs, and observability to ship on-device AI with a focus on speed and privacy.
Runs on-device inference for iOS, macOS, Android and edge devices via a cross-platform SDK stack (Swift, Kotlin, React Native, Flutter) that exposes a single API across platforms. The MetalRT engine uses hand-written kernels, operator fusion, and unified memory optimization to achieve high throughput (e.g., 668 tok/s LLM decode on Apple Silicon) and low latency (e.g., 101 ms STT). The product comprises an inference engine (MetalRT), developer SDKs across languages, and a control plane for fleet management including OTA model updates and routing, enabling production deployment with device-level execution and analytics.
Who it’s for: mobile and edge developers, AI teams building on-device/inference workloads for iOS, macOS, Android and edge devices
backed by Y Combinator; public research and SDK releases; GitHub presence; ongoing publications and benchmarks
Former Intuit engineer building RunAnywhere, the infrastructure layer for deploying fast, private, multimodal AI on-device at scale. Deep background in mobile SDKs, platform tooling, and developer products, including systems used by 50M+ active users. Previously founded products across consumer discovery, context management, agentic documentation, and mobile testing, and now focused on making on-device AI production-ready across mobile, edge, and embedded devices.
AmazonCo-founder & CTO of RunAnywhere (W26). Built MetalRT: the first complete multi-modal inference engine for Apple Silicon. Custom Metal GPU kernels that pushed on-device voice AI from 900ms to ~110ms. Ex-Amazon EC2 Spot ($100M+ ARR), Ex-Microsoft Azure. Peer-reviewed researcher.
One SDK + Control Plane to Deploy, Route, Update, and Observe On-Device AI – Offline-First, Hybrid-Smart, Fleet-Managed
RunAnywhere provides a full-stack on-device AI infrastructure with a single SDK and a control plane to manage multi-engine models, delivery, and policies. It enables offline-first, hybrid routing for on-device AI across iOS, Android, React Native, and Flutter, targeting developers needing reliable local inference with observed performance and OTA updates.
Enabling Model-Hardware Co-Design at Scale
Deployable Physical AI on Any Edge Device.
