MacPaw and Liquid AI announced a long-term partnership on August 5, 2026 to build local AI for the Mac. The assistant is designed to run directly on the device, without sending data to the cloud, and the technology will eventually reach developers through the Setapp marketplace.
Key Takeaways
- The partnership pairs Liquid Foundation Models (LFMs) with two MacPaw technologies — the Elix inference engine and the Mnemos memory layer.
- The first product will be a local version of the Eney assistant, running on Apple silicon.
- Results of the collaboration are expected in the second half of 2026.
- Setapp has more than 150,000 paying users, and AI operations will be billed with credits based on task complexity.
- Liquid AI already serves customers such as Mercedes-Benz, Insilico Medicine, and Shopify.
A full AI stack on the device
MacPaw, a Kyiv-based app developer with offices in Boston and across the EU, is building a complete local AI stack for macOS together with Liquid AI. Liquid AI, an MIT spinout founded in 2023, supplies its Liquid Foundation Models family optimized for assistance on the Mac. MacPaw adds two of its own components: Elix, an engine that runs the models directly on the device, and Mnemos, a memory layer that keeps context across interactions.
The on-device AI stack has three layers:
Ramin Hasani, co-founder and CEO of Liquid AI, explains the edge of this approach through hardware-specific architecture.
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, with benefits like privacy and security.
Ramin Hasani, co-founder and CEO of Liquid AI.
Eney as the first test
The first product to use the partnership will be Eney, MacPaw's assistant for macOS. Its local version is meant to combine three things in one product — processing on Apple silicon, persistent memory, and task execution directly inside the system. According to both companies, that combination has not been available on the Mac before. When a task needs more power, the assistant is designed to reach for cloud models, for example from Google.
MacPaw also says the stack will let models improve over time using user data, and that agentic workflows can run offline.
Oleksandr Kosovan, CEO of MacPaw, frames the goal as building a full AI stack for the Mac where personal data stays on the device.
We believe intelligence should live where people work — private by design, fast by default.
Oleksandr Kosovan, CEO of MacPaw.
Local versus cloud
Betting on on-device processing is a response to two dominant market models. On one side are assistants that run mostly in the cloud and send every query to servers. On the other are Apple's local models built into the operating system. MacPaw aims in between — its own stack, where data stays on the device but with a path to the cloud when a task requires it.
The billing model differs too. Instead of a flat subscription, AI operations in Setapp are paid for with credits tied to task complexity. For Liquid AI it is another edge deployment after customers in automotive, biotech, and e-commerce. For MacPaw it is a first step from a single app toward an AI platform for the whole Setapp ecosystem.
Why it matters
Local AI is moving from novelty to real product strategy. If MacPaw ships an assistant that works offline, remembers context, and executes tasks inside the system, it shows that privacy and convenience need not conflict. The more important part is the second half of the plan — opening the stack to Setapp developers. That shifts the company from app maker to AI infrastructure provider for the Mac, a space Apple largely controls today. A credit model, rather than a flat subscription, also shows AI compute cost starting to shape consumer software pricing.
What's next
- Results of the collaboration, led by the local version of Eney, are expected in the second half of 2026 according to both companies.
- MacPaw plans to open the stack — models, the Elix engine, and Mnemos memory — to thousands of Mac developers through Setapp.
- AI billing in Setapp is set to run on a credit model tied to task complexity.
