ai.diy
The AI workspace that never leaves your browser.
The story
Overview
ai.diy is a browser-owned AI workspace that prioritizes user ownership and privacy. Unlike traditional AI platforms that store keys and data on servers, ai.diy keeps your workspace state, chats, files, and provider credentials entirely in your browser using IndexedDB. The application acts as a relay layer that forwards requests to your chosen AI provider without storing sensitive credentials server-side. It supports 20+ AI providers including OpenAI, Claude, Gemini, Groq, Ollama, and custom OpenAI-compatible endpoints, allowing you to switch models mid-conversation without losing context. The workspace includes Canvas for creating presentations and documents, Python execution via Pyodide, npm package support, and local knowledge-base search with on-device embeddings. Users can self-host the MIT-licensed application on their own Node.js server or Docker infrastructure, or use the public demo. ai.diy is designed for developers, researchers, and privacy-conscious users who want full control over their AI interactions and data.
Key features
Browser-Owned Storage
All chats, files, Canvas artifacts, memory, and settings persist locally in IndexedDB with no server-side data storage required.
20+ AI Provider Support
Connect to OpenAI, Claude, Gemini, Groq, Ollama, and 15+ other cloud and local models, switching between them mid-conversation.
Bring Your Own Keys
Provider credentials stay in your browser and are relayed per request, eliminating the need for persistent server-side API keys.
Canvas Artifacts
Create presentations, documents, code, and files directly within the workspace using Canvas, Python, npm packages, and browser tools.
Local Knowledge Base
Search your notes and PDFs using on-device embeddings while keeping the index local to your browser.
Self-Hosted Deployment
Deploy the MIT-licensed open-source application on your own Node.js server, Docker, or Vercel infrastructure in one command.
Use cases
- 1
Privacy-Conscious Developers
Build AI-powered applications while maintaining full control over API keys and data without relying on third-party servers.
- 2
Research and Analysis
Conduct research with local knowledge bases, document RAG, and multi-model exploration without uploading sensitive documents to cloud services.
- 3
Content Creation
Generate presentations, documents, and code artifacts using Canvas and multiple AI models while keeping all work local.
- 4
Self-Hosted Teams
Deploy ai.diy on internal infrastructure for teams that require complete data sovereignty and control over AI provider integrations.
FAQ
Are my AI provider keys stored on the ai.diy server?
No. Provider keys are kept in your browser and relayed per request to the provider you select. No server-side LLM credentials are required or stored.
Where does ai.diy store chats and documents?
Chats, files, Canvas artifacts, memory, and knowledge-base chunks persist in your browser through IndexedDB and localStorage. Optional backups to S3, WebDAV, or Google Drive are client-side features you enable yourself.
Which AI providers work with ai.diy?
ai.diy supports 20+ integrations including OpenAI, Anthropic, Google Gemini, Groq, Ollama, Mistral, Hugging Face, Amazon Bedrock, Azure, Vertex, and custom OpenAI-compatible endpoints.
Can I self-host ai.diy?
Yes. Run the production build on a standard Node.js server, use Docker Compose, or deploy to Vercel. The server acts as a request relay and does not need provider API keys in environment variables.
Does ai.diy replace my AI provider or pay for model usage?
No. ai.diy is the workspace layer. You bring your own authorized provider keys or connect local models. Provider pricing and usage charges remain your responsibility.
Tech stack & tags
Feedback & Discussion
Discussion
The founder is asking for feedback
What would make you choose ai.diy as your everyday AI workspace instead of ChatGPT, Claude, or another AI tool?
Sign in to leave a comment or rating.
No comments yet. Be the first.
