holaOS

holaOS is an open-source desktop app that serves as a local-first AI agent for workplace management. It focuses on learning a user's working context to automate tasks and manage project workflows without requiring the user to manually provide background information for every session. The software is available as a desktop application for macOS, Linux, and WSL.

Users connect the agent to their existing professional tools via OAuth to create a working memory of their current state. The application replaces the need for terminal-based interactions by providing a graphical user interface where users can manage tasks, inspect files, and view outputs in a shared workspace. It supports various state-of-the-art models through a single account system, removing the need to manage multiple API keys.

Key features

• OAuth integration with over 100 workplace tools including GitHub, Slack, and Jira
• Local-first memory storage using Markdown files and SQLite vec for RAG retrieval
• Session context compression to maintain reasoning coherence in long-running tasks
• Browser profile integration for interacting with web-based dashboards and apps
• Parallel execution via hidden subagents coordinated by an orchestrator agent
• Local workspace for storing plans, drafts, and generated files outside of chat logs
• Automatic background fetching of relevant signals from connected tools
• User-controlled memory management allowing for manual editing and deletion

The system architecture is built on Electron and TypeScript, utilizing a memory mechanism inspired by the LLM wiki workflow. It transforms raw data from integrations into a compressed knowledge base, which ensures that the agent does not have to search through every raw email or ticket from scratch. This approach is designed for professionals who require a durable continuity of work across multiple days and complex projects.

holaOS is a production-grade working agent that prioritizes local data ownership and business-level memory management.