MacPaw Supercharges App Store Devs with Liquid AI's On-Device Inference Breakthrough

Ukraine-based app developer MacPaw has announced a strategic partnership with Liquid AI, aiming to integrate locally hosted AI models into its product ecosystem and, subsequently, make this advanced tech stack available to a broader developer community. This collaboration underscores MacPaw's commitment to enhancing its offerings with cutting-edge artificial intelligence capabilities.

A significant part of this initiative involves preparing MacPaw's popular subscription-based app store, SetApp, for a new generation of AI applications. The company plans to introduce credit-based plans for users, allowing them to engage with AI operations based on allocated credits and the complexity of the tasks performed. SetApp currently boasts over 150,000 paying users, indicating a substantial platform for the rollout of these new AI features.

Central to MacPaw's AI strategy is the development of its AI assistant, Eney, which was first unveiled last year. The current focus is on building a locally hosted version of Eney. For this endeavor, MacPaw has enlisted Liquid AI to develop critical components: an on-device inference system named Elix and a robust local memory system. These systems are designed to ensure that AI processing can occur directly on the user's device, offering significant advantages.

Ramin Hasani, co-founder and CEO of Liquid AI, elaborated on their architectural approach, stating, “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, with benefits like privacy and security.” This bespoke hardware-tailored architecture is key to achieving optimal performance and user data protection.

Oleksandr Kosovan, CEO of MacPaw, further emphasized the user benefits of locally hosted AI models, noting that they will provide users with the crucial ability to run assistants and various agentic workflows offline. This offline capability enhances user accessibility and convenience, freeing them from constant internet connectivity requirements for basic AI functions.

While companies like Apple already offer their own local models to developers, Hasani asserts that Liquid AI's models distinguish themselves by prioritizing performance across a diverse range of capabilities. Furthermore, Liquid AI is actively constructing a customization stack around its models. Hasani explained, “We are also building a customization stack around models. This means that with user input, the models can use the data and improve. We want our models to be adaptable and become more intelligent over time.” This adaptive learning mechanism will allow the AI to evolve and become more tailored to individual user needs over time.

Once the local processing architecture is firmly established with Liquid AI, MacPaw intends to democratize this technology by making it accessible to other developers. This will enable them to integrate on-device inference directly into their own applications. Kosovan also revealed that the platform will extend its utility by offering access to other cloud models from major providers like Google, positioning SetApp as a comprehensive, one-stop shop for developers seeking diverse AI capabilities.