Run an agent on your own computer, work with local files, use local models when you can, and choose cloud providers only when you want them.

OpenYak is built for people who want an AI agent that lives on their machine instead of behind another hosted workspace account.

• No OpenYak account. Install the app, choose a local model or bring your own provider key, and start working without login, billing, seats, or recharge flows.
• Local-first agent runtime. Files, conversations, memory, generated artifacts, tool permissions, and workflow state stay on your device.
• Work from your actual files. Attach DOCX, XLSX, PPTX, PDFs, CSVs, and local project context, then ask for briefs, tables, follow-ups, plans, and reusable artifacts.
• Keep the workflow in one thread. Start with analysis, continue into a RACI, ask for a follow-up email, and preserve context across long conversations.
• Choose your model boundary. Run local models through Rapid-MLX or Ollama, or opt into direct BYOK calls to OpenRouter, OpenAI, Anthropic, Google, and other providers.
• Use your desktop agent from another device. Remote access lets you scan a QR code and send tasks to your computer through a secure tunnel.

OpenYak can turn a dense memo into a structured brief that is ready for a manager, team update, or follow-up email.

Use spreadsheets as working inputs, not screenshots. Ask for budget variance, forecast risks, owner-level action items, and meeting-ready talking points.

OpenYak can synthesize several files in the same thread and open a right-side artifact panel for reusable briefs, plans, diagrams, and structured outputs.

Real office work rarely fits in one message. OpenYak is designed for follow-ups, revisions, and long conversations where the important context needs to remain available.

Download the latest release or visit open-yak.com/download.

Linux users can also read LINUX.md for requirements and troubleshooting.

1. Install OpenYak for your platform.
2. Choose where inference runs: local Rapid-MLX/Ollama for offline work, or a BYOK cloud provider when you want hosted models.
3. Start a new conversation and attach a real file.
4. Ask for a deliverable, not just a summary: brief, action plan, RACI, email, table, or artifact.
5. Review the result in the chat and artifact panel, then continue in the same thread.

Example prompt:

Please read the files I uploaded and turn them into a concise team brief.
Start with three key takeaways, then list risks, owners, and next actions.
Finally, write a follow-up email I can send to the team directly.

• Rapid-MLX: Apple Silicon macOS users can start and switch curated MLX models from Settings. OpenYak connects to Rapid-MLX's OpenAI-compatible API on localhost.
• Ollama: Run any model available through Ollama. Local models are auto-detected and can be used without an internet connection.
• Custom local endpoints: Point OpenYak at a local OpenAI-compatible server when you run your own model stack.

Cloud and subscription paths are optional. OpenYak does not provide hosted model accounts and does not proxy model traffic; requests go directly from your desktop to the provider you configure.

• File understanding: office docs, spreadsheets, slide decks, PDFs, CSVs, local folders, and generated artifacts.
• Artifact workspace: reusable Markdown briefs, tables, diagrams, checklists, and structured outputs.
• Tool execution: read, write, rename, organize, and automate files with user-controlled permissions.
• Long-context work: continue from analysis to planning to follow-up without starting over.
• Remote access: connect from mobile through QR code and Cloudflare Tunnel.
• Automations: schedule recurring cleanup, reporting, and file workflows.
• Privacy controls: local storage, no OpenYak account, BYOK provider access, and local model support.

Tech Stack: Tauri v2, Rust, Next.js 15, FastAPI, SQLite

Monorepo Structure:

desktop-tauri/    Rust desktop shell and system integration
frontend/         Next.js chat UI, settings, artifacts, and SSE streaming
backend/          FastAPI agent engine, tool execution, LLM streaming, storage

Quick Start:

npm run dev:all

This starts the backend on port 8000 and the frontend on port 3000. For deeper setup notes, see frontend/README.md and backend/README.md.

• Questions and Discussions: GitHub Discussions
• Bug Reports: GitHub Issues
• Contributing: CONTRIBUTING.md

MIT