Ukraine-based software developer MacPaw has announced a strategic partnership with Liquid AI, a pioneer in efficient artificial intelligence models, to integrate locally hosted AI capabilities across its product ecosystem. This collaboration marks a significant shift for MacPaw as it seeks to transition from cloud-dependent AI features to on-device inference, prioritizing user privacy, reduced latency, and offline functionality. The partnership will initially focus on enhancing MacPaw’s AI assistant, Eney, and will eventually extend a specialized technology stack to the thousands of third-party developers who distribute their software via Setapp, MacPaw’s popular subscription-based app store.
The move comes at a time when the broader technology industry is pivoting toward "Edge AI"—the practice of running complex machine learning models directly on consumer hardware rather than in centralized data centers. By leveraging Liquid AI’s specialized architecture, MacPaw aims to provide a more responsive and secure user experience, distinguishing its offerings from competitors that rely solely on internet-connected LLMs (Large Language Models).
The Technical Foundation: Liquid AI and the Elix System
At the heart of this partnership is the development of an on-device inference system known as Elix. Unlike traditional AI models that are often resource-intensive and designed for massive server farms, Liquid AI’s models are built on a different architectural foundation. Liquid AI, a startup spun out of the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL), has gained industry attention for its "Liquid Neural Networks." These models are designed to be more computationally efficient, requiring less memory and processing power while maintaining high levels of intelligence.
Ramin Hasani, co-founder and CEO of Liquid AI, noted that the collaboration involves selecting and tailoring model architectures specifically for the hardware they will inhabit. This optimization ensures that the AI can run natively on Mac and iOS devices without draining battery life or requiring constant cloud connectivity. The Elix system will also incorporate a "local memory system," allowing the AI to learn from user interactions and retain context over time without that data ever leaving the physical device.
This customization stack is a critical differentiator. Hasani emphasized that these models are designed to be adaptable, improving as they process user input. By focusing on performance across various capabilities, Liquid AI intends to offer a level of specialization that generic cloud models often lack.
Enhancing the Eney AI Assistant
MacPaw’s primary vehicle for this new technology is Eney, an AI assistant the company unveiled in late 2023. Named after the protagonist of the Ukrainian epic "Eneida," Eney was designed to help users navigate their digital workflows more efficiently. By integrating Liquid AI’s local inference capabilities, MacPaw plans to evolve Eney into a more robust, "agentic" assistant.
An agentic workflow refers to an AI’s ability to not only answer questions but to execute multi-step tasks autonomously. For example, a locally hosted Eney could potentially organize files, summarize documents, or automate system settings even when the user is offline. Oleksandr Kosovan, CEO of MacPaw, highlighted that local hosting is the key to unlocking these capabilities. By removing the "middleman" of the cloud, the assistant becomes faster and more reliable, particularly for users in environments with unstable internet connections or those handling sensitive corporate data.
Strategic Shift for Setapp and the Developer Ecosystem
Beyond its own internal tools, MacPaw is positioning itself as a platform provider for the broader developer community. Setapp, which currently boasts over 150,000 paying subscribers and offers a curated selection of more than 240 apps, is being prepared for a new era of AI-centric software.
Once the local processing architecture developed with Liquid AI is finalized and stabilized, MacPaw intends to open the tech stack to external developers. This would allow independent creators to integrate sophisticated on-device AI into their own apps without having to build the underlying infrastructure from scratch.
To support this, MacPaw is also evolving the economic model of Setapp. The company is experimenting with credit-based pricing for AI operations. Under this system, users would receive a certain number of credits as part of their subscription, which they could spend on various AI tasks. The "cost" in credits would vary based on the complexity of the task—for instance, a simple text correction might cost one credit, while a complex image generation or a deep data analysis might cost more. This approach addresses the high computational costs associated with AI while providing a predictable revenue stream for developers.
Furthermore, Kosovan stated that the platform will act as a "one-stop shop" for developers. While the focus is on Liquid AI’s local models, the platform will also provide streamlined access to cloud-based models from providers like Google. This hybrid approach allows developers to choose the best tool for their specific use case, whether it requires the privacy of local processing or the massive scale of the cloud.
Market Context and the Competitive Landscape
The partnership between MacPaw and Liquid AI arrives in a highly competitive landscape dominated by tech giants. Apple recently introduced "Apple Intelligence," its own suite of on-device AI features powered by its proprietary silicon. While Apple provides local models to developers via its Core ML and Apple Intelligence APIs, Liquid AI and MacPaw argue that there is room for alternative architectures that offer different performance profiles and deeper customization.
Liquid AI recently raised $250 million in funding, valuing the company at significant levels and signaling strong investor confidence in its alternative to the Transformer-based models used by OpenAI and Google. By partnering with MacPaw, Liquid AI gains a massive real-world testing ground with a sophisticated user base that values high-quality software and privacy.
The broader market for AI is also shifting. According to industry analysts, the "AI PC" market is expected to grow significantly, with a forecasted 60% of all PC shipments by 2027 being AI-capable. MacPaw’s investment in local inference aligns with this hardware trend, ensuring that its software can take full advantage of the Neural Processing Units (NPUs) being integrated into the latest generations of Mac and mobile hardware.
Privacy, Security, and Offline Resilience
One of the most compelling arguments for the MacPaw-Liquid AI partnership is the enhancement of user privacy. In a standard cloud-based AI interaction, user data is sent to a remote server, processed, and then returned. Even with encryption, this creates a potential point of vulnerability. Locally hosted models eliminate this data transfer entirely.
For MacPaw, a company headquartered in Ukraine, there is also a practical element of resilience. The ongoing conflict in the region has highlighted the importance of software that can function without a constant link to global infrastructure. By enabling offline agentic workflows, MacPaw ensures that its tools remain functional for users in any circumstance.
Implications for the Future of macOS and iOS Development
The collaboration between MacPaw and Liquid AI represents a significant milestone in the democratizing of advanced AI tools. For years, only the largest tech firms had the resources to develop and deploy high-performance AI. By creating a reusable stack for on-device inference, MacPaw is lowering the barrier to entry for smaller, independent developers.
This could lead to a surge in specialized, "quiet" AI—features that work in the background to improve user experience without the fanfare or high latency of current chatbots. From better predictive text and automated photo tagging to complex system optimization, the potential applications of the Elix system are vast.
As MacPaw continues to refine its credit-based pricing and developer tools, the industry will be watching closely to see if this model can become a sustainable alternative to the current ad-supported or high-cost subscription models that dominate the AI landscape. For the 150,000 users of Setapp, the immediate benefit will be a smarter, faster, and more private suite of tools that work seamlessly across their devices.
About the Companies
MacPaw is a software company based in Kyiv, Ukraine, known for creating popular macOS and iOS utilities including CleanMyMac X, Gemini, and the Setapp subscription service. Founded by Oleksandr Kosovan, the company has become a staple of the Apple ecosystem, focusing on user-centric design and system optimization.
Liquid AI is a Boston-based AI research and development company. Spun out of MIT, the firm specializes in creating a new class of AI models that are dynamic, efficient, and capable of running on edge devices. Led by CEO Ramin Hasani, the company is at the forefront of developing alternatives to traditional generative AI architectures, focusing on scalability and real-time adaptability.
