Gerty

Product Introduction

1. Definition: Gerty is a private, local AI assistant application for macOS. Technically, it is a unified interface and automation layer that automatically selects, downloads, and manages multiple open-source large language models (LLMs) and runs them entirely on-device on Apple silicon Macs.
2. Core Value Proposition: It exists to provide a fully private, offline-capable alternative to cloud-based AI assistants like ChatGPT. Its primary value is complete data sovereignty—ensuring no user data, queries, documents, or voice inputs ever leave the user's Mac, addressing critical privacy, confidentiality, and vendor lock-in concerns.

Main Features

1. Automated Local Model Management: Gerty automatically evaluates the user's Mac hardware (Apple silicon chip type and available RAM) and selects the best-performing open-source models (like Gemma, Qwen, Llama, Mistral) for different tasks (chat, writing, coding, vision). It handles the entire setup process, including downloading and configuring the models, eliminating the need for manual terminal commands or technical knowledge of model files and formats.
2. Deep System and Document Integration: The assistant works contextually with native macOS apps and user data. It can read and answer questions based on content from Mail, Calendar, Contacts, Notes, and files (including PDFs, with OCR for scanned documents). It can execute actions within these apps and generate new documents (PDF, Word, Excel), requesting user approval before making any changes. This integration is facilitated through local, on-device connectors.
3. Fully Offline, Voice-First Operation: All AI processing occurs on the Mac's Neural Engine and CPU/GPU. This enables a fully functional voice assistant accessible from the macOS menu bar notch, supporting voice queries and local neural text-to-speech responses in 31 languages, all without an internet connection. A local, OpenAI-compatible API is also provided for integration with other coding tools like IDEs.

Problems Solved

1. Pain Point: The privacy risk and data governance complexity of sending sensitive personal, professional, or confidential information to third-party cloud AI services. It also solves the technical barrier to entry for running powerful local AI models.
2. Target Audience: Professionals handling sensitive data (lawyers, doctors, consultants, journalists), privacy-conscious individuals, developers seeking a local AI API, and macOS power users wanting integrated, context-aware assistance without subscriptions.
3. Use Cases: Analyzing confidential legal or financial documents locally; summarizing private email threads or calendar events; generating and editing code with full IP control; getting voice answers offline while traveling; preparing meeting notes from internal documents without data leakage.

Unique Advantages

1. Strengths & Limitations (Pros & Cons):

   • Pros: Unmatched data privacy and security (zero data transmission); one-time purchase license (CHF 49) for 3 Macs, offering significant long-term savings versus subscriptions; seamless, automated setup for local AI; deep and respectful integration with macOS apps and files.
   • Cons: Requires specific hardware (Apple silicon Mac with 16GB+ RAM for best performance); model capabilities are limited by local hardware and the selected open-source models, potentially lagging behind the latest massive cloud models in reasoning or knowledge breadth; initial model downloads are large (multi-gigabyte).

2. Key Alternatives & Differentiation:

   • ChatGPT/Cloud AI Subscriptions: Differentiates on privacy (local vs. cloud), cost (one-time vs. recurring), and offline functionality. Gerty sacrifices the raw power and constant updates of GPT-4 for absolute data control.
   • Ollama/LM Studio: These are technical "model runners." Gerty differentiates by being a polished end-user "assistant." It adds automated model selection, a unified chat interface, system integrations, voice features, and document interaction on top of the local inference engine, removing the need for manual configuration.
   • Other Local AI Apps (e.g., MacGPT): Often rely on connecting to cloud APIs or are single-feature tools. Gerty differentiates with its comprehensive, all-in-one approach combining automated multi-model local inference, document analysis, and system automation into a single cohesive product.

Frequently Asked Questions (FAQ)

1. Is Gerty a good ChatGPT alternative? Yes, for users prioritizing privacy, offline use, and cost predictability. It is a direct alternative for day-to-day tasks like writing, summarization, coding help, and document Q&A, with the critical distinction that all processing occurs locally on your Mac without any data leaving your device.
2. What are the system requirements for Gerty? Gerty requires a Mac with Apple silicon (M1, M2, M3, or M4 chip) and macOS 26 (Sequoia) or later. While it will run on base models, a Mac with 16 GB or more of unified memory is recommended for optimal performance with larger, more capable models.
3. Can Gerty read and edit PDF files? Yes. Gerty can read the text content of PDFs, including performing OCR on scanned documents. It can also edit PDFs in place, preserving the original layout, and create new PDF files from scratch, always showing you a preview and asking for confirmation before saving any changes.
4. How does the one-time license work? A single purchase license for Gerty costs CHF 49 and is valid for use on up to three Macs owned by the licensee. This includes all future updates within the same major version line, with no subscription or recurring fees.
5. Does Gerty work completely without an internet connection? Yes, once the AI models are downloaded, all core functionalities—chat, document processing, voice interaction, and app integration—work entirely offline. An internet connection is only needed for initial model downloads, optional web search, update checks, and for enabling certain optional cloud-based connectors.