MacPaw Taps Liquid AI for On-Device Inference, Opening Its App Store to AI Developers

MacPaw, the Ukrainian developer known for Mac utilities like CleanMyMac, is doubling down on on-device AI. The company has partnered with Liquid AI to develop locally hosted inference and memory systems for its upcoming AI assistant Eney, with plans to eventually offer the same infrastructure to developers building for its SetApp subscription store.

The collaboration centers on a custom on-device inference system called Elix, along with a local memory system. Liquid AI’s CEO, Ramin Hasani, emphasized that their models are designed with hardware efficiency in mind: "Before training our models, we select an architecture that is different and tailored to the hardware. That allows us to really have the most efficient version of intelligence that runs directly on the device."

For MacPaw, the payoff is clear. CEO Oleksandr Kosovan says locally hosted models will enable users to run assistants and agentic workflows offline, a major differentiator. While Apple already offers its own local models for developers, Hasani argues Liquid AI’s models are tuned for performance across different capabilities and include a customization stack that lets models improve with user input.

SetApp, which already has over 150,000 paying users, is set to become a hub for AI apps. MacPaw is experimenting with credit-based pricing, letting users perform AI operations based on credits and task complexity. Once the local architecture is locked in, MacPaw plans to make the tech stack available to developers, positioning SetApp as a one-stop shop that also provides access to cloud models from providers like Google.

This move could be a quiet but significant turn for the Mac software ecosystem. On-device inference offers tangible benefits: privacy, security, and low latency, all without a subscription to a cloud AI service — a key selling point for users weary of sending data to remote servers. For developers, having a ready-made local inference stack in a store with a subscription model could lower the barrier to building AI-powered apps that work offline. The question is whether Liquid AI's architecture will deliver the performance and adaptability needed to compete with Apple's on-device models, which are already deeply integrated into the platform. If MacPaw succeeds, SetApp could emerge as a real alternative for developers who want more control over their AI stack.