MacPaw taps Liquid AI to offer on-device inference to devs building for its app store
TechCrunch Ivan Mehta
MacPaw just teamed up with Liquid AI to run AI models right on your device, not in the cloud. That means offline assistants and eventually letting other developers plug into the same tech through MacPaw's app store.
Based on reporting by TechCrunch, Ivan Mehta — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
MacPaw, the Ukraine-based company behind the Setapp app store, is betting its future on AI that lives on your device instead of some distant server. The company has partnered with Liquid AI to build the guts of a locally hosted version of Eney, the AI assistant it introduced last year. The project has a name now too: Elix, an on-device inference system paired with a local memory setup that Liquid AI is building specifically for this job.
Liquid AI's co-founder and CEO Ramin Hasani explained the thinking to TechCrunch this way: before any training happens, the team picks an architecture tailored to the actual hardware it'll run on. That choice, he says, is what lets the intelligence run efficiently right on the device, with privacy and security as a natural side effect of never sending your data anywhere. MacPaw CEO Oleksandr Kosovan framed the payoff in practical terms too, saying local models mean users can run assistants and agentic workflows even when they're offline.
This isn't happening in a vacuum. Apple already hands developers its own local models to build with. Hasani doesn't dodge that comparison; he just argues Liquid AI's models are built around performance across different capabilities, and that the company is layering a customization stack on top so models can absorb user input and get smarter over time rather than staying static.
The bigger play here is Setapp itself, which already counts more than 150,000 paying users. MacPaw wants to lean harder into AI apps on that platform, and once the local processing architecture with Liquid AI is locked down, the plan is to open it up to outside developers so they can bake on-device inference into their own apps. Kosovan says the platform will also plug into cloud models from companies like Google, positioning Setapp as a one-stop shop rather than a walled garden.
MacPaw is also testing credit-based pricing for AI features on Setapp, where the number of operations a user gets depends on how many credits they have and how demanding the task is. It's a small detail, but it hints at how the company is trying to figure out the economics of AI apps before it throws the doors open to third-party developers.
My take — AI-written commentary, not fact-checked reporting
Going local isn't just a privacy pitch, it's MacPaw hedging against becoming another company renting its core product from a cloud AI vendor. Betting on a smaller player like Liquid AI instead of just leaning on Apple's built-in models is the more interesting call here, and it only pays off if the on-device performance actually holds up once real developers start building on it.
Read more about this at: TechCrunch
