Philip Durbin 🚀 (Oct 15 2026 at 19:57): Philip Durbin 🚀 (Oct 15 2026 at 19:57):Hi! Here at Harvard Dataverse we're starting to use a tool we developed with Firebrand (firebrand.ai) for AI-powered spam filtering and metadata validation.
You can see some of the history of what we've been up to in a GitHub issue.
We are planning some webinars for the community. Please feel free to ask any questions that we should cover!
Augustus Saunders (Oct 19 2026 at 20:18):Hi everyone, I’m Augustus Saunders, founder and CTO of Firebrand. Thanks to Sonia and the team for making space for us here.
We’ve been working with Harvard Dataverse on ReviewIQ to help curators spend less time screening out spam and submissions that aren’t datasets. It’s been running in production since August, and Harvard estimates it’s giving its curation lead about eight hours back each week.
ReviewIQ uses deposit metadata and file listings to screen new submissions against the repository’s curation rules. It provides written reasons for its results, and curators make the final call on anything it holds for review. Deposits that pass the AI review continue and publish as they normally would.
It connects through Dataverse’s existing validation hooks, with no core changes, and runs in the repository’s own cloud. We can tune the screening rules to reflect how your team works.
We’re making ReviewIQ available to other installations as a commercial product. If this kind of screening is taking up more of your team’s time than you’d like, we’d be happy to share a short walkthrough and compare notes on where it could help.
Last updated: Nov 02 2026 at 07:01 UTC