MacPaw, the Ukrainian software company best known for CleanMyMac, has announced a partnership with Liquid AI to bring on-device inference to developers building apps for its alternative Mac app store. The deal centers on a local version of MacPaw's AI assistant Eney, rebuilt using Liquid AI's efficient model architecture to run entirely on-device — no cloud round-trip required.

Why On-Device, Why Now

The timing isn't accidental. With Apple opening the door to alternative app marketplaces under regulatory pressure — particularly in the EU — MacPaw's Setapp Mobile and broader app distribution ambitions are gaining real stakes. Offering on-device AI inference as a developer platform feature is a meaningful differentiator: it addresses privacy concerns, reduces latency, and eliminates per-query API costs that can erode margins for indie developers.

Liquid AI, spun out of MIT in 2023, has built its reputation on Liquid Foundation Models (LFMs) — a non-transformer architecture designed for efficiency at inference time, particularly on edge and resource-constrained hardware. That makes them a logical fit for MacPaw's goal of running capable AI on consumer Mac hardware without leaning on cloud infrastructure.

What Developers Actually Get

The practical implication for developers building within MacPaw's ecosystem is access to inference capabilities that:

  • Run locally on the user's device, keeping data off external servers
  • Leverage Liquid AI's model efficiency to work within Mac hardware constraints
  • Are surfaced through Eney, MacPaw's existing AI assistant layer

MacPaw appears to be positioning Eney not just as an end-user product but as a platform-level AI layer — something developers can build against rather than around.

Liquid AI's Growing Distribution Play

For Liquid AI, the MacPaw deal represents a continued push into embedded and on-device deployment channels. The company has been quietly building partnerships that get its models into production environments rather than competing head-on with OpenAI or Anthropic for API traffic. Winning distribution through platform partners — where their model efficiency story is most compelling — looks like a deliberate go-to-market strategy.

This contrasts with the dominant cloud-inference playbook. Most AI model providers monetize through hosted API calls. Liquid AI's architecture lets it compete on a different axis: total cost of ownership for the developer or platform operator, not just benchmark scores.

Implications for Founders and Indie Developers

For startup founders and independent developers in the Apple ecosystem, this partnership points to a near-term future where on-device AI is a realistic baseline expectation, not a premium engineering challenge.

A few things worth watching:

  • Cost structures shift — if inference is local, developers don't pay per-call cloud fees, which changes unit economics for AI-native features
  • Privacy as a feature — on-device processing is increasingly a selling point with enterprise and regulated-industry buyers, not just a technical detail
  • Platform lock-in dynamics — building on MacPaw's ecosystem means betting on its traction as a viable alternative to the App Store, which remains uncertain outside the EU

MacPaw hasn't disclosed the commercial terms of the Liquid AI arrangement or a specific timeline for developer-facing availability. But the direction is clear: on-device inference is moving from research curiosity to platform infrastructure, and the companies that establish distribution now will have a meaningful head start.