ShogunAI

Product Introduction

1. Definition: ShogunAI is a private, local-first AI memory assistant and execution agent for macOS. It is a desktop application that functions as a personal Artificial General Intelligence (AGI) layer, operating directly on a user's Apple Silicon Mac.
2. Core Value Proposition: It exists to close the gap between AI-generated answers and finished work. Unlike cloud-based chatbots that start each session with zero context, ShogunAI builds a persistent, private memory of a user's work across connected tools (email, calendar, documents, chat) and uses that context to autonomously execute tasks—from drafting replies to filing action items—while maintaining strict user approval gates for any external communication.

Main Features

1. Private, On-Device Memory Layer: ShogunAI continuously captures and indexes work context (text, source, timestamps) from authorized applications like Gmail, Notion, Calendar, and Slack. It uses macOS's native accessibility layer for text capture and, optionally, encrypted local screenshots for non-readable content. This creates a searchable, offline-available knowledge graph of people, projects, and commitments stored encrypted on the local disk.
2. Context-Aware Execution Layer: This is the core automation engine. It reads the built memory, identifies open loops (e.g., a follow-up promised in an email), and acts to close them inside the user's actual tools. Actions are governed by a three-tier autonomy model: reversible background tasks run automatically, drafts are created for review, and any action that would reach another person (sending an email, posting a message) requires explicit user approval.
3. Unified Model Context & BYOK (Bring Your Own Key): The application provides a single, structured memory profile to any AI model or interface the user employs, including its own notch panel, CLI, or tools connected via MCP (Model Context Protocol). Users can run it using their existing OpenAI or Anthropic subscription quotas or supply their own API keys, allowing control over costs and model provider without losing their local memory context.

Problems Solved

1. Pain Point: The "cold start" problem of generative AI. Every interaction with standard chatbots like ChatGPT or Claude requires manually re-pasting context, recent decisions, and open threads, wasting time and limiting usefulness.
2. Target Audience: Knowledge workers and professionals whose effectiveness depends on context spread across multiple tools—such as Founders, Executives, Product Managers, Consultants, and Engineers. It is built for individuals who manage complex projects, communications, and commitments.
3. Use Cases: Automatically generating and filing meeting minutes with action items; drafting context-aware email replies inline with a keyboard shortcut; performing unified semantic search across all connected work apps; providing a daily brief of overnight updates and pending commitments; and translating live conversations while preserving a text transcript.

Unique Advantages

1. Differentiation: Unlike cloud-based AI assistants or simple meeting recorders, ShogunAI combines persistent, private memory with autonomous execution. Competitors typically offer either memory/recall or automation, but not a tightly integrated, local-first system that does both. It acts as a harness around the entire work stack, not a separate chat window.
2. Key Innovation: Its architecture is fundamentally local-first. Capture, indexing, storage, and search occur on the user's Mac. Data only leaves the device in minimal, auditable slices when necessary for AI processing (e.g., sending a thread for summarization). This ensures privacy, eliminates reliance on a cloud server that could be breached, and enables offline functionality for memory recall.

Frequently Asked Questions (FAQ)

1. How does ShogunAI's privacy model work compared to cloud AI? All primary data capture, indexing, and storage happen locally on your Mac. Your raw work memory never leaves your device. Only the specific context needed for a discrete task (e.g., an email thread to draft a reply) is sent to your chosen AI provider, and all such events are logged in a local audit trail. Audio from meetings is never written to disk.
2. Can I use ShogunAI with my existing OpenAI or Claude subscription? Yes. ShogunAI is designed to work with the "assistant" plans from major providers, consuming your existing API credits. You can also bring your own API key (BYOK). This gives you full control over the model provider, rate limits, and costs.
3. What specific actions can ShogunAI automate? It executes tasks within your connected tools, such as drafting emails in Gmail, creating or updating pages in Notion, creating calendar events, and drafting messages in Slack. The key differentiator is that these drafts or updates are created in situ with full context, and any action destined for another person is gated by a mandatory approval step.
4. Is ShogunAI just a better meeting transcription tool? No. While it can transcribe and summarize meetings, that is one input to its memory system. Its core value is connecting meeting outcomes to the rest of your work state and acting on them—for example, by automatically creating a follow-up task in your project management tool from a verbal commitment made during the call.
5. Why is it only available for macOS? The product is deeply integrated with native macOS frameworks like the Accessibility API for context capture, the system keychain for secure connections, and Apple Silicon for efficient on-device processing. Porting it to Windows or Linux would require rebuilding these core architectural components from the ground up.