Hi everyone, I'm Flynn, and I built Arbor because I know how painful literature review is firsthand.
During my PhD in Biomedical Engineering, I spent countless hours searching for papers, cross-referencing findings, and trying to piece together the state of research on a topic. It was slow, fragmented, but a vital part of the research process.
Arbor automates that entire workflow. You drop in a question, and AI agents:
- Decompose it into focused sub-inquiries
- Search across arXiv and Semantic Scholar in parallel
- Screen papers for relevance and extract key findings
- Synthesize everything into a structured overview
The entire pipeline streams in real-time as an interactive knowledge graph, so you can watch the research unfold and explore any branch.
The stack: React + TypeScript frontend, FastAPI backend, DeepSeek V3 for decomposition and synthesis, Gemini 2.0 Flash for paper screening, and GPT-4o-mini for content moderation.
I'd love feedback on the research quality, the graph UX, or anything else. What would make this more useful for your workflow?
