MacPaw AI gives Mac products the on-device runtime to think, remember, and act.
Native-Swift inference, memory, and agents - fast and private by design.
Already in our products. Opening to yours.
Everyone who joins the MacPaw platform: developers, product teams, and partners.
The full on-device AI stack for the Mac - describe intent in Swift, and a shared service runs it on an engine tuned for Liquid Foundation Models.
Learn about ElixDescribe your intent in declarative Swift, and the compiler parallelizes the graph - open source under Apache 2.0.
One process serves every app - managing resources and offloading the LLM when idle to free the user's RAM.
Best-in-class performance for Liquid Foundation Models - MLX-based, tuned per chip.
MacPaw's on-device AI memory layer - a Swift package that gives any assistant a persistent, searchable memory of the user, private by default.
Learn about MnemosA governed knowledge graph - people, projects, and facts as first-class entities, contradictions resolved, every fact with citations.
What you told it, what's written in your data, what it observed - three trust levels, kept apart, every fact traceable to its source.
Local models by default, secrets in an encrypted vault, sensitive data redacted before it reaches any model.
Every top AI model in one place - one OpenAI-compatible endpoint for your apps, with no keys to manage and no billing to build.
Visit Setapp AI GatewayClaude, GPT, Gemini, DeepSeek, and 10+ more - switch models by changing a single parameter.
Dynamic routing sends simple queries to cheaper models, and repeats are cached - lower cost, same UX.
Setapp validates subscriptions and AI-credit balances for you - ship AI features without a payment stack.
The shipping side of MacPaw AI - flagship products running on the platform.
EneyEney is a local-first AI assistant for your Mac. It gets things done right where you are - no round trips for what your Mac can handle itself.
Local features run on Elix, the Apple-native agentic AI runtime, with Mnemos as long-term memory. Fast, private, and in real time.
Visit
CleanMy®Phone EngineThis engine uses machine learning to organize, declutter, and filter your media library.
It finds blurry shots, duplicates, and screenshots that no longer spark joy, giving you back storage space without losing anything important.
Visit
SetappSetapp is a curated subscription of best-in-class Mac and iOS apps - one membership, hundreds of tools.
Its AI assistant matches the task you describe to the right app, and Setapp AI brings top AI models into member apps - one subscription covers it all.
VisitMacPaw Research open-sources models, datasets, and tooling as they mature. A selection of what is shipping today.
BLIP-based captioner for macOS icons - drives accessibility labelling across our pipelines.
View on Hugging Face ↗YOLOv11-Large fine-tuned for macOS UI element detection - the backbone behind Screen2AX.
View on Hugging Face ↗Sibling YOLOv11-Large model - detects UI groups like toolbars and tab bars.
View on Hugging Face ↗UI Parsing and Accessibility Dataset - 352 macOS app screens with accessibility trees.
View on Hugging Face ↗Structured metadata for thousands of Mac App Store apps.
View on Hugging Face ↗Automated macOS UI exploration trajectories for GUIrilla-See.
View on Hugging Face ↗Tools, ports, and engines MacPaw publishes for the community. Built in the open, on GitHub.
SwiftUI-like DSL for composing AI pipelines in Swift.
View on GitHub ↗Community-driven Swift package for the OpenAI API.
View on GitHub ↗Swift port of Gliner2 - entity recognition on the Mac.
View on GitHub ↗Fast-dLLM diffusion-LLM inference, ported to MLX.
View on GitHub ↗Extract structured accessibility trees from any Mac app.
View on GitHub ↗Automated agent for exploring and testing macOS GUI flows.
View on GitHub ↗An open, vendor-neutral memory format and protocol for AI.
View on GitHub ↗Swift implementation of the portable memory protocol.
View on GitHub ↗Contributing to the field, not racing it. Selected AI research papers from MacPaw Research.
We collaborate, we don't compete. From foundation models to academic research - partnerships are how MacPaw AI ships faster than it could alone.
Our foundation-model partner - Liquid AI's LFMs are efficient by architecture. We built Elix to run them leaner still on Apple silicon.
Our voice AI partner - studio-grade voice cloning and real-time TTS, trusted by Lucasfilm and Sony. Integrated into Mac voice features via Elix.
Our partner for purpose-built small language models that replace expensive LLM calls - fueling the training pipeline behind our Adapter Store.
University collaborations driving co-authored work, internships, and our long-term talent pipeline.
Joint research and internships in on-device inference.
Joint applied-AI research and a steady ML talent pipeline.
Joint research, internships, and a talent pipeline.
Home of our "Bilka Space" AI Lab and an ML scholarship.
Working on a Mac product that needs on-device intelligence? Want to partner, license a model, or talk roadmap? Drop us a line - we read every email.
Email the team → [email protected]Selected coverage of the AI strategy, partnerships, and product launches.
