Mirai Secures $10M to Build On-Device AI Layer
AI startup Mirai has secured $10 million in seed funding to develop an on-device AI “capability layer.” The company's goal is to make local AI inference more accessible for both consumer and enterprise application developers. The investment will be used to expand the engineering team and accelerate product development.
- The founding team consists of Dima Shevts and Alexey Moiseenkov, the entrepreneurs behind the viral consumer AI apps Reface, which had over 200 million downloads, and Prisma. - The seed round was led by Uncork Capital and includes angel investments from notable figures such as Snowflake co-founder Marcin Żukowski, ElevenLabs co-founder Mati Staniszewski, and Coinbase board member Gokul Rajaram. - Mirai's proprietary inference engine, built in the Rust programming language and initially optimized for Apple Silicon, has shown performance improvements of up to 37% faster generation on certain models compared to alternative runtimes. - The company's technology is designed to solve the complexity of running AI models locally, which typically requires deep expertise in memory management and hardware-aware execution that most app developers lack. - Mirai's system allows for a hybrid approach, enabling developers to route some AI tasks to the device for speed and privacy while keeping other parts of their workflow in the cloud. - Moving AI processing from the cloud to devices addresses key issues like high cloud costs, interaction latency, and data privacy, while also enabling apps to function offline. - With the new capital, the company plans to expand its support beyond Apple Silicon to other hardware and extend its capabilities to include voice, vision, and multimodal AI workloads.
