MacPaw taps Liquid AI to offer on-device inference to devs building for its app store
Ukrainian software developer MacPaw has partnered with AI firm Liquid AI to build locally hosted, on-device AI models, with plans to eventually offer this technology to developers building apps for its SetApp store. The move signals a push towards privacy-focused, offline-capable AI tools, positioning MacPaw's platform as a rival to Apple's own on-device AI offerings for developers.
Under the deal, Liquid AI is developing an on-device inference system called Elix and a local memory system to power a locally hosted version of MacPaw's AI assistant, Eney, first unveiled last year. MacPaw chief executive Oleksandr Kosovan said this would let users run AI assistants and agentic workflows offline, while also giving SetApp's more than 150,000 paying subscribers access to cloud-based models from firms such as Google. The company is also trialling credit-based pricing for AI features on the app store, letting users pay based on the number and complexity of AI operations performed.
In short
- MacPaw partners with Liquid AI for on-device, offline AI models
- Aims to offer the tech stack to developers on its SetApp store
- SetApp is also testing credit-based pricing for AI features
Steelman: both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The strongest case for on-device AI emphasises user privacy, autonomy and security. When AI models run locally, personal data never leaves the device, eliminating reliance on distant cloud infrastructure and reducing exposure to data breaches. This approach also enables offline functionality and reduces latency, giving users faster responses whilst maintaining control over their information. For many, privacy and data sovereignty are fundamental rights that should not require trusting large technology companies with sensitive data.
The case against
The strongest case for cloud-based AI highlights the substantial capabilities advantage that centralised models provide. Cloud AI systems can be vastly more sophisticated and continuously improved, delivering superior results that on-device models cannot match due to hardware constraints. For developers and users seeking the best possible AI experiences, the trade-off of cloud processing is worthwhile, and privacy concerns can be addressed through robust regulations, transparent practices and user consent rather than by limiting technological capability.
Explained simply
MacPaw is a Ukrainian software company that makes apps for Apple devices. Setapp is its subscription service, where people can access a collection of apps; developers can build apps for the platform, and subscribers pay to use them.
AI models process information to produce answers or carry out tasks. They can run on remote computers, often called the cloud, or directly on a user’s device; local processing can work without an internet connection and may keep more information on the device.
MacPaw’s assistant, Eney, is intended to use Liquid AI’s technology to run locally, with cloud-based models also available. The plans matter to developers because MacPaw hopes to make this kind of AI available for apps on Setapp, alongside a pricing approach based on how much AI processing users do.