Discourse AI adds decision model integration to Discourse Workflows, letting site administrators evaluate community posts against custom classification questions for automated moderation, notifications, and tagging.
Discourse AI integrates specialized decision models, such as TypeSafe Jev, directly into Discourse Workflows through a Classify step. Instead of running full generative text prompts, this feature scores text and images against targeted questions. Intended uses include flagging contentious conversations for human review, routing notifications to specific team members, and organizing topics with tags. The decision models tab is currently restricted to sites running the Workflows feature, and the shared announcement provides configuration screenshots rather than formal performance evaluations.
How you can use it
On a Discourse site with Workflows, configure a decision model in the AI administration area. Then add a Classify step that asks about a post or image and connects the answer to a forum action. Start with staff notifications or tags so you can check mistakes before trusting automatic moderation.
