MacPaw taps Liquid AI to offer on-device inference to devs building for its app store

Ukraine-based app developer MacPaw has teamed up with Liquid AI to enhance its products with locally hosted AI models, aiming to eventually offer the technology stack to other developers. MacPaw is also preparing its SetApp app store for AI applications, introducing credit-based plans for users. The company is presently developing its AI assistant, Eney, which it introduced last year, and intends to create a locally hosted version, facilitated by Liquid AI's on-device inference system, Elix, and local memory system.

Liquid AI's co-founder and CEO, Ramin Hasani, explained to TechCrunch that they choose an architecture tailored to the hardware during model training, enabling the most efficient AI model running directly on the device, thereby ensuring privacy and security. MacPaw's CEO, Oleksandr Kosovan, stated that locally hosted AI models will allow users to run assistants and agentic workflows offline.

Interestingly, Apple already offers its own local models to developers. However, Hasani emphasizes that Liquid AI's models prioritize performance for various capabilities. Their team is also constructing a customization stack around these models, enabling users to input data, which the models utilize to improve over time. MacPaw intends to concentrate on AI apps for its subscription-based app store, SetApp, which currently boasts over 150,000 paying users.

Once the local processing architecture is secured with Liquid AI, MacPaw aims to make the technology accessible to other developers, providing them with on-device inference capabilities for their apps. Additionally, MacPaw intends to offer access to other cloud models, such as those from Google, functioning as a comprehensive platform for developers.

MacPaw is also exploring credit-based pricing for its app store, allowing users to execute a certain number of AI operations based on their credits and the complexity of the task.