ai.diy

什麼是 ai.diy？

ai.diy is a browser-owned AI workspace built for bring-your-own-key chat, research, and artifact creation. It keeps chats, files, knowledge, settings, and other workspace state in the browser while routing provider requests through a self-hosted relay.

The product is designed for people who want one workspace for cloud and local models, rather than separate apps for each provider. It supports switching between 20+ providers, including cloud services and local endpoints such as Ollama and LM Studio, while keeping the browser as the primary storage layer for threads, Canvas, memory, and knowledge.

ai.diy 能做什麼？

Browser-local workspace state

Chats, files, Canvas artifacts, memory, knowledge, usage events, and settings persist in the browser through IndexedDB and localStorage.

Bring-your-own-key provider access

Provider credentials stay in browser storage and are relayed per request, so the server does not need to store persistent provider keys.

20+ cloud and local integrations

The workspace can connect to many providers, including OpenAI, Anthropic, Gemini, Grok, OpenRouter, DeepSeek, Ollama, LM Studio, Bedrock, Azure, Vertex, Together, Mistral, Hugging Face, and custom OpenAI-compatible endpoints.

Model switching within a thread

Users can change models mid-thread without moving their workspace or copying context into another app.

Built-in tools and artifact creation

The workspace includes Canvas, Python, npm packages, and browser tools for making presentations, documents, code, and other files.

Self-hosting support

The app can be run as a production Node.js build or with Docker Compose, and the relay does not require provider keys in server environment variables.

使用情境

“Personal AI workspace with local persistence”

Users who want chats and notes to remain in the browser can keep their work in one place without depending on a hosted provider workspace.

“Comparing cloud and local models”

Developers and power users can test a local endpoint like Ollama or LM Studio alongside cloud providers in the same interface.

“Private document and note search”

Users can search PDFs and notes with on-device embeddings while keeping the index local, then decide whether retrieved context should be sent to a chosen model.

“Building outputs in one thread”

Teams or individuals can use the workspace to generate documents, code, presentations, and files with tools next to the conversation.

“Self-hosted deployments”

Organizations or individuals who need to control infrastructure and network boundaries can run the app themselves and keep the relay separate from provider credentials.

常見問題

What is ai.diy?

ai.diy is an open-source, self-hosted AI workspace for chat, research, tools, Canvas artifacts, memory, and local knowledge.

Where are provider keys stored?

Provider keys are kept in the browser and relayed per request. The server is not used as a persistent key store.

Can I self-host ai.diy?

Yes. The product can run as a production build on Node.js or through Docker Compose.

Does ai.diy work with local models?

Yes. It supports local providers such as Ollama and LM Studio, as well as other OpenAI-compatible endpoints.

What does the pricing page show?

The pricing route returns a not-found page, so no pricing details are available from the provided sources.