AI moderation
AI moderation that explains its work.
ReplyTidy combines contextual AI, your own rules, author controls, and local spam checks so every new Threads reply receives a visible, recoverable decision.
ai moderation / current
More than a word list
AI rules can interpret the meaning and intent of a reply, helping recognize variations of harassment, scams, repetitive promotions, and audience-specific behavior that exact keyword filters miss.
Your policy chooses the action
AI identifies likely matches; it does not invent your moderation policy. You choose whether a category or custom rule should allow the reply, hide it, or send it to review.
Safety categories and thresholds
Configurable categories cover harassment, hate, threats, violence, sexual content, and illicit or violent content. Each category can use its own action and confidence threshold.
Plain-language, keyword, and hybrid rules
Describe a pattern in ordinary language, match an exact phrase, or combine both approaches. Hybrid rules are useful when a recurring phrase matters but context should determine whether it truly matches.
Visible reasons, not black-box outcomes
The decision history records whether a category, custom rule, author control, or spam signal influenced the outcome. Reviewers can inspect that evidence before approving, hiding, restoring, or marking a false positive.
Human control remains final
Moderation models can make mistakes. Preview mode shows what would be hidden before changing the live conversation, uncertain replies can wait for review, and hidden replies can be restored.
See the moderation flow