Rivul AI – Built semantic search and MCP server on OpenAlex data
Arjun Prabhulal
Hi all,
I wanted to share Rivul AI (https://rivul.ai), a free Multimodal AI Research Workspace I build and maintain myself, using the OpenAlex dataset.
Its paper search covers 300M+ records from an OpenAlex snapshot (June 2026). You can search in plain language and it matches by meaning, not just keywords. For example, “why do neural networks forget earlier tasks when trained on new ones” finds the catastrophic forgetting literature. In the workspace you can also add PDFs (including scans and images), ask questions across them with your own AI key, and listen to papers read aloud.
The same search is available through a remote MCP server, so AI assistants can use it directly, with filters for field, topic, year, citation count and open access. There are setup guides for Claude Code, OpenAI Codex, Cursor, Google Antigravity and Muse Code, and it works with any MCP client that supports Streamable HTTP. You can connect by signing in through the browser (no key to copy) or with an MCP key. The tools are search_papers and get_paper_details, and there’s a REST API too.
Sign-in is with Google, and there are no paid plans.
I’d love to hear from researchers, librarians and anyone who works with OpenAlex data: would searching it by meaning, or from an AI assistant, be useful to you, and what would you want it to return? Any feedback is appreciated, especially where the metadata looks wrong.
Thanks to the OpenAlex team for making this data openly available; Rivul AI search is built on it.
Regards,
Arjun
Parthasarathi Mukhopadhyay
--
You received this message because you are subscribed to the Google Groups "OpenAlex Community" group.
To unsubscribe from this group and stop receiving emails from it, send an email to openalex-commun...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/openalex-community/CAM1%3DYcy1iwsTyKF%3DvKocuoc11VVHUJh_7nPvcr%3DCmZTD91RMNw%40mail.gmail.com.