Mirai raises $10M to optimize on-device AI performance

Mirai, a London-based startup, has secured $10 million in seed funding led by Uncork Capital to develop an inference engine—the software component that executes artificial intelligence models directly on hardware. Founded last year, the startup’s 14-person technical team is working to improve how models run on consumer devices like phones and laptops, shifting workloads away from centralized cloud servers. The seed funding will support the team as they build out their framework to optimize on-device performance.

The company’s engine, built in the programming language Rust, can increase a model’s generation speed by up to 37%. Mirai aims to simplify the deployment process for developers by offering a Stripe-like integration experience. Shvets, co-founder of Mirai, explained that the company wants to provide developers with an integration experience requiring only eight lines of code, allowing them to connect to the platform and immediately run tasks like summarization or classification. While the startup’s current framework is optimized for Apple Silicon, the team plans to bring its engine to Android devices in the future to expand its compatibility across different mobile operating systems.

Co-founders Dima Shvets and Alexey Moiseenkov bring deep consumer-tech experience to the venture, having previously co-founded the face-swapping app Reface and the AI-filter app Prisma, respectively. Both founders have experience in building scalable consumer applications, which informs their focus on the user experience and hardware constraints of mobile devices. Shvets noted that during the current generative AI boom, “When we met together in London, we started to chat about technology, and we realized that within the hype of GenAI and more AI adoption, everybody speaks about cloud, about servers, about AGI coming. But the missing piece is on-device [AI] for consumer hardware”

Andy McLoughlin, managing partner at Uncork Capital, supported this view, noting that the current venture capital trend of funding massive cloud inference costs is unsustainable. He explained that as businesses begin to focus on underlying economics, model makers will increasingly seek to run portions of their inference workloads at the edge, positioning Mirai to capture this demand.

Why it matters

The shift toward on-device AI is driven by the need to reduce the high costs of cloud-based inference, positioning edge-optimized engines as a critical infrastructure layer for future consumer applications.