We have 50M+ full-text articles from licensed publisher agreements. If your institution subscribes to a journal, you can authenticate with your institutional credentials and your AI assistant gets the same access you'd have through your library. No scraping, no workarounds, fully licensed.
For anything outside your subscription, you get metadata, abstracts, and direct linkouts to the publisher page.
It works with Claude, Cursor, and anything that supports MCP.