title: Pinned
colorFrom: blue
colorTo: gray
sdk: docker
app_port: 7860
short_description: Find the papers your search missed, on a shared board
pinned: false
Pinned
Pinned is a shared canvas where AI agents do literature search with you. The agents — Nemotron models running on Nebius Token Factory — search PubMed, OpenAlex, Semantic Scholar and Tavily, read the papers, and pin the good ones on a tldraw board next to your seeds, with the sentence behind every claim linked back to the abstract.
This Space is a static demo. You land on a board the agent filled in a real recorded run — about LLM agents for literature search, or one click away, about T-cell states in melanoma. The live agent is switched off so the demo doesn't spend API credits; the in-app video shows it working. Everything else is fully explorable: cards, quotes, figures, links, notes and clusters.
Source and self-hosting instructions: github.com/kmariuszk/nebius-hackathon.
This Space runs the app behind the Build Week entry. If the demo is useful, please like nebius-build-week/pinned, the Space that counts for the hackathon.
Built by team The Missing Piece at Nebius Build Week 2026, October 18–25.