Graph your way from Papers to Possibilities

Papers are added by DOI. Open a graph's Paper Registry, paste a DOI, and Axy fills in the title, authors, year and abstract from CrossRef. You then attach those papers as evidence to the nodes and links they support. Axy does not store PDFs, and there is no bulk import from Zotero, Mendeley or a reference file yet. Axy also does not read your papers and build the graph for you. You build the map, and the papers back up each claim on it.

A private graph is visible only to your organisation and the people you invite into it. A public graph can be found and read by anyone with an Axy account, and anyone can propose changes to it, but only your organisation's admins decide what gets merged. Public graphs are how a community builds one shared map of a field in the open. The two kinds count against your plan's limits separately, and you can switch a graph between private and public in its settings.

Nobody writes straight into a graph. Everything you change in Build mode is saved into a submission of your own, which you open for review when it is ready. Reviewers see exactly what would change, discuss it in threads, and approve it or ask for revisions. An admin of the graph's organisation then merges it, after Axy checks that the graph has not changed underneath it. That is how a lab keeps one trustworthy graph while many people work on it, and how outsiders can safely propose changes to a public graph.

Discover mode lets you ask your graph questions in plain language, such as "what connects the gut microbiome to Parkinson's in this map?" The agent searches the graph, runs read-only queries, and draws the relevant part of the graph on a canvas beside the chat. It can never edit the graph. Discover mode runs through Claude Code on your own computer, so you need Claude Code installed and signed in with your own Anthropic account. Axy shows you the install command the first time you open the tab. Everything else in Axy works without it. You can pick which Claude model answers, and Axy never stores your Discover conversations. They stay in your own Claude Code session history.

Reference managers like Zotero or Mendeley organize papers — flat lists of PDFs and citations. Axy organizes knowledge. It maps the concepts inside those papers into a structured graph that captures what a researcher has learned, not just what they have read. The graph is shared, collaborative, and cumulative, so a lab's understanding compounds instead of retiring with each person.

It stays. That is the core problem Axy solves. Today, when a postdoc leaves, their mental model of the field leaves with them. With Axy, the knowledge graph they built is a permanent, shared asset the lab inherits — the next person does not start from scratch, they start from where the last person left off.