Mirai: Optimizing AI Models on Mobile Devices

Mirai, founded by the creators of Reface and Prisma, has raised 10 million dollars to optimize how AI models run on devices such as smartphones and laptops.

Mirai: Optimizing AI Models on Mobile Devices
Mirai logo, a startup focused on optimizing AI models on mobile devices.

In today's AI landscape, attention is often focused on expanding cloud capabilities and building huge data centers to run models. However, a group of developers has decided to tackle a fundamental challenge: improving how AI models run directly on users' devices, such as smartphones and laptops. This is the goal of Mirai, a London-based startup founded by Dima Shvets and Alexey Moiseenkov, both with solid experience building scalable consumer applications. Shvets is known for co-founding the face-swapping app Reface, while Moiseenkov was CEO and co-founder of the AI filter app Prisma. (techcrunch.com)

The Problem with Running AI Models on Devices

Although many modern devices include dedicated AI chips, most AI applications still rely on the cloud to process models. This approach presents significant challenges, including high latency, operating costs and privacy concerns. Running models directly on users' devices can reduce latency, improve privacy and lower the costs associated with cloud processing. (trymirai.com)

Mirai's Proposed Solution

Mirai has developed a proprietary inference engine optimized for Apple Silicon, designed to improve how AI models run directly on devices. The engine is built in Rust and promises significant gains in model generation speed, up to 37% faster than traditional inference methods. Mirai also offers an SDK that lets developers integrate the engine into their apps easily, with just a few lines of code and without complex system-level work. (trymirai.com)

Implications for the Sector and Future Outlook

Mirai's focus on optimizing on-device AI execution represents an important step toward widespread adoption of real-time, low-latency AI. As concerns about data privacy and security grow, processing AI models locally offers significant advantages. In addition, the ability to run complex models directly on devices can open new opportunities for advanced applications in fields such as healthcare, security and entertainment. (trymirai.com)

Conclusion

  • Innovation in on-device AI: Mirai is tackling the challenge of improving how AI models run directly on devices, reducing dependence on the cloud.
  • Key advantages: On-device processing offers lower latency, better privacy and lower operating costs.
  • Future outlook: Mirai's approach could pave the way for new advanced applications of real-time AI on mobile devices and laptops.

For more details on Mirai and its solutions, you can visit the company's official website. (trymirai.com)