Our sentiment classifier reads every new comment and hides the obviously negative or off-brand ones below the post. Your audience never sees them; you can always review and restore.
💬 “Loved the recipe — making it tonight!” → kept (positive) 💬 “Bhai 50% off kab aayega” → kept (neutral) 💬 “This brand is a scam, don't trust them” → hidden (negative) 💬 “DM me for cheap followers 🔥💯” → hidden (off-brand) 💬 Slur targeting commenter → hidden (toxic)
Same model the AI-replies feature uses for sentiment routing. Per-post toggle lets you turn shielding off on posts where you want raw feedback (e.g. a beta launch).
When you tag a brand partner, the shield turns on automatically. The sponsor's audit reviewer never sees the troll comment buried at position 47.
/inbox shows the shielded comments with a 'Restore' button. Wrong call by the AI? Restore in one tap, and the model learns from the override on your account.
Set per-account or per-post sensitivity — strict (hide on lean-negative) for sponsored, balanced for the main feed, loose for personal vlogs where banter is the point.
Compliance-ready audit log: comment text, classifier score, model version, timestamp, and the post it was on. Export via REST API for legal reviews.
A hidden comment is treated as non-existent for automation purposes — no DM is sent, no lead is scored. Trolls don't get free AI replies.
We hard-classify targeted harassment, slurs, and threats — those hide regardless of your sensitivity setting. Your audience and team are protected.
Free Starter plan, 1,000 DMs/month, 60-second Instagram connect. Upgrade to Pro+ when you want this feature on.