Build a personal AI tutoring workspace that remembers your materials and learning history
By connecting read documents, past conversations, and learning progress in a single space, the AI understands the context and answers questions. DeepTutor is a personalized learning workspace that remembers the user's knowledge base and learning history, growing with them. The latest version, v1.7.2, adds desktop downloads, selected text speech synthesis, and a Korean interface preview, improving accessibility.
One runtime, diverse learning modes
DeepTutor integrates various learning modes, including chat, quiz, research, visualization, problem solving, mastery path, and reading, into a single runtime. Each mode shares session context while maintaining a workflow suited to its purpose. The home, learning, and resource areas change according to the active workspace, while partner and personalization settings are available across the entire account. Users can create workspaces for specific topics or projects to manage learning materials and track progress.
Grounded answers and inspectable memory
The core of this system is grounded answers based on user materials and inspectable memory. You can build a version-controlled knowledge library using various indexing engines such as LlamaIndex, PageIndex, GraphRAG, and LightRAG. Memory is divided into L1 traces, L2 facts, and L3 summaries, making personalization visible and editable, and auditable through evidence links. Additionally, you can connect external agents like Claude Code, Codex, and Grok CLI as partners to consult in real time or retrieve and utilize past conversations.
Installation and execution
DeepTutor can be installed as a desktop app, via Docker, or as a Python package. The desktop build bundles Python and Node.js, allowing immediate execution without using the terminal, with installers provided for Mac and Windows. Using Docker, you can run the web app with a single docker run command and preserve settings and local data by specifying a data volume. To install with Python, you need Python 3.11~3.14 and Node.js 20.9+, and you can install both the web app and CLI together with the pip install -U "deeptutor[full]" command.
Pre-use checks
DeepTutor requires a Python 3.11+ and Node.js 20.9+ environment, and when using Docker, port 3782 must be open. When upgrading from versions prior to 1.6.15, you must remove the existing deeptutor-cli package, back up the runtime home, and then install the new package. Additionally, certain features may require installing additional modules or configuring external API keys, so it is recommended to check the features you intend to use and install the necessary dependencies. The project is recruiting community contributors, so if you wish to participate in maintenance, you can refer to the contribution guide.
