Local Deep Research

AI-powered research assistant that searches 10+ sources — arXiv, PubMed, web, and your private documents — then generates comprehensive, cited reports. Supports local LLMs (via Ollama) and cloud providers (OpenAI, Anthropic, Google).

First Launch

1. Open your domain — the web UI loads at the root URL
2. Configure your LLM — set your preferred model provider in the settings (default uses Ollama if available)
3. Start a research query — type a question and let the assistant search, analyze, and compile a report
4. Review the report — results include citations, source links, and confidence scores

Key Features

• Multi-source research: arXiv, PubMed, Wikipedia, web search, and local documents
• Supports local LLMs (Ollama) and cloud APIs (OpenAI, Anthropic, Google)
• Per-user encrypted SQLite database — all data stays on your server
• Iterative deep research with configurable search depth
• Export reports in multiple formats

Environment Variables

Set LDR_LLM_PROVIDER and LDR_LLM_API_KEY to configure your preferred LLM backend. See the documentation for all available settings.

Persistent Data

Research data and configuration are stored in /data (backed by a Zeabur volume).

License

MIT — GitHub