MacPaw Liquid AI partnership targets on-device AI for Mac developers
MacPaw and Liquid AI are building a local AI stack for Eney first, with developer access planned only after the architecture is finalized.
By Jordan Bell · Startups & Deals Reporter
· 3 min read
The MacPaw Liquid AI partnership aims to put more AI processing directly on Macs, starting with MacPaw’s Eney assistant. For users, the potential appeal is an assistant designed to work locally rather than sending every request to a remote data center. For developers, however, the proposed on-device technology is a future offering, not a product available today.
MacPaw and Liquid AI announced the collaboration on Aug. 5 to co-develop a local AI stack for Mac, according to TechCrunch and Forbes. The companies plan to pair Liquid AI’s models with MacPaw’s Elix inference engine and Mnemos memory and context layer. Inference is the step where a trained AI model processes a request and produces an answer or action.
Eney is intended to be the first product to use the stack. Forbes reported that the companies expect core assistant work to run locally through Elix on Apple silicon, while cloud models could remain available for tasks where they are a better fit. MacPaw has not announced a release date for the local version of Eney.
What does the MacPaw Liquid AI partnership mean for developers?
MacPaw wants to offer the local-processing architecture to developers after it has finalized it, CEO Oleksandr Kosovan told TechCrunch. Kosovan also said the platform is expected to provide access to cloud models, including models from Google. That would create a proposed route for app makers to use both on-device inference and cloud AI through one platform.
The scope and timing remain unsettled. MacPaw has not announced an SDK, hardware requirements, pricing for developers, or confirmed that access will be limited to Setapp developers. Liquid AI’s separate LEAP platform, which supports Android and iOS, should not be read as a specification for the MacPaw project.
MacPaw says local models could support privacy, quicker responses and offline operation. Kosovan told TechCrunch that users could run assistants and agentic workflows without an internet connection. Those are stated product aims rather than independently measured results.
Setapp is preparing for AI apps
The partnership arrives as MacPaw shifts more attention to AI apps in Setapp, its subscription app store. TechCrunch reported that Setapp has more than 150,000 paying users. The company is also testing credit-based pricing, in which the number of AI operations available to a user depends on credits and the complexity of a task.
MacPaw’s own July update said Setapp’s AI Router and AI Gateway are intended to help developers build and operate AI-powered apps. Those tools are separate from the planned local AI architecture with Liquid AI.
Apple is already part of the competitive backdrop. In June 2025, Apple said its Foundation Models framework would let developers access an on-device Apple Intelligence model for private, offline experiences, with inference free of charge. MacPaw’s effort therefore adds another planned option in an ecosystem that already has Apple’s native model path, rather than replacing it.
This story draws on original reporting from TechCrunch.
