Mirai Raises $10M for On-Device AI

- Mirai was founded by Dima Shvets and Alexey Moiseenkov, the minds behind the popular AI photo apps Reface and Prisma. Their experience with the challenges of running complex AI on consumer devices, like battery drain and the need for cloud processing, directly led to the creation of Mirai. - The $10 million seed round was led by Uncork Capital and included angel investments from notable figures in the tech industry, such as the co-founder of ElevenLabs, a co-founder of Snowflake, and a former Apple Distinguished Engineer who worked on MLX. - Mirai's proprietary inference engine is built in Rust and is initially optimized for Apple Silicon, claiming performance improvements of up to 37% faster generation and 59% faster prefill on certain models compared to alternatives like MLX and llama.cpp. - The company's strategy is to address the high costs and latency associated with cloud-based AI by enabling more processing to happen directly on user devices, a trend accelerated by major tech companies like Apple, Google, and Microsoft, who are increasingly integrating on-device AI into their flagship products. - Mirai aims to simplify the process for developers to run AI models locally, offering an SDK that requires just a few lines of code to integrate, removing the need for deep expertise in areas like memory management and kernel optimization. - Beyond its initial focus on Apple Silicon and text-based models, Mirai plans to expand its support to include Android devices and other modalities like voice and vision. They are also developing an orchestration layer to manage hybrid workflows between the device and the cloud. - The on-device AI space is becoming increasingly competitive, with other startups like Modal Labs and Quadric also securing significant funding to tackle similar challenges in AI inference optimization.