Keel is a local-first AI assistant for Mac and Windows. It stores your data as plain markdown on your disk, supports multiple AI models (Claude, GPT, OpenRouter, Ollama), and offers features like auto-capture, wiki bases, meeting transcription, and task management. No telemetry, no account, no lock-in.
Free
How to use Keel?
Download and install Keel on your Mac or Windows machine. It creates a local workspace folder (~/Keel) where all your notes, projects, and history are stored as markdown files. You can chat with your chosen AI model, which indexes your workspace for context. Use commands like /create-kb for wikis, record meetings for automatic transcription and summary, or set up scheduled jobs for daily digests. Your data never leaves your machine unless you explicitly share it.
Keel 's Core Features
Markdown workspace: All data stored as plain markdown files in a folder you control, editable in any text editor and easily backed up.
Multi-model support: Swap between Claude, GPT, OpenRouter, or local Ollama models anytime without losing context or data.
Auto-capture: Automatically saves important decisions, facts, and tasks from conversations back into your workspace as structured markdown.
Meeting transcription: Record or import audio, transcribe locally with Whisper, and generate structured summaries with decisions and action items.
Wiki bases: Turn any project folder into a queryable knowledge base using markdown and PDFs, with automatic syncing as files change.
Task management: First-class to-dos with due dates, project assignments, and desktop notifications for reminders.
Voice input: Speak instead of type using local Whisper or OpenAI's API for hands-free interaction.
Keel 's Use Cases
Professionals who want a private, local AI assistant that organizes daily tasks and notes without cloud dependency.
Project managers needing to build and maintain knowledge bases from project folders for quick reference.
Remote workers who attend many meetings and want automatic transcription and structured summaries saved locally.
Developers who prefer open-source, self-hosted tools and want to experiment with different AI models.
Writers and researchers who need a smart workspace that captures ideas and organizes them into searchable wikis.
Anyone concerned about data privacy, wanting an AI that runs entirely offline and doesn't track usage.
