AI Research Assistant¶
A local-first research pipeline that retrieves academic papers from multiple scholarly APIs, ranks and clusters them with embeddings, synthesizes cross-paper insights, and exports reports in several formats. Uses Ollama by default for fully local LLM inference, with optional OpenAI and Anthropic providers.
Built with Python 3.13, pydantic-ai, sentence-transformers, and async I/O.
Feature list, requirements, and RAM guidance: README.
Default quality profile
Synthesis and query expansion default to heuristic mode (llm_enabled: false). Reports are fast but template-driven. Enable LLM features for richer cross-paper analysis — see Heuristic vs LLM.
Quick links¶
| Section | Description |
|---|---|
| Installation | Pipenv, Python 3.13, dependencies |
| Quick Start | First query and auto-setup flow |
| CLI Reference | Flags, batch vs interactive mode |
| CLI vs API | Execution paths and provider divergence |
| Configuration | Env vars, YAML, and precedence |
| Known Issues | Research quality analysis and fix backlog |
| Architecture | End-to-end pipeline overview |
| Canonical sources | Where copy-paste commands live (link, don't duplicate) |
Install and run: README Quick Start.