MacPaw, a Ukraine-based app developer, has joined forces with Liquid AI to bring locally hosted AI models to its products, with plans to eventually open up the technology stack to other developers. The company is also gearing up its SetApp subscription service to accommodate AI apps, introducing a credit-based system for users.

Currently, MacPaw is developing its AI assistant, Eney, which was first revealed last year. The next step is to create a locally hosted version of this assistant, and to make that happen, the company has brought in Liquid AI to build an on-device inference engine called Elix, along with a local memory system. "Before training our models, we select an architecture that is different and tailored to the hardware," MacPaw’s CEO, Oleksandr Kosovan, explained. He also noted that locally hosted AI models would give users the freedom to run assistants and agentic workflows even without an internet connection.

Apple already offers its own local models to developers, but Liquid AI’s co-founder, Ramin Hasani, argues that their models are designed with a focus on performance across various capabilities. "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," Hasani said.

On the app store front, MacPaw is doubling down on AI applications for SetApp, which currently boasts over 150,000 paying subscribers. Once the local processing architecture with Liquid AI is finalized, MacPaw intends to make the technology available to developers, allowing them to integrate on-device inference into their own apps. Kosovan mentioned that the platform would also serve as a gateway to other cloud-based models, such as those from Google, effectively becoming a one-stop shop for developers. In the meantime, MacPaw is already testing a credit-based pricing model for the store, where users can perform AI operations based on the credits they hold and the complexity of each task.