Try it on a real conversation.
Empty channel. Send a message below, try an example to see how context changes the call.
Two stages. One decision.
Gate
Every message passes a fast, free filter: OpenAI moderation plus a multilingual bad-word net that survives obfuscation. The clean ~90% stop here. Nothing is spent.
Judge
Only flagged messages reach the AI judge. It reads the surrounding conversation and rates intent on a 1 to 5 scale (sarcasm and banter at the bottom, real threats at the top) with a written reason.
Act or ask
Each severity maps to actions you configure: delete, warn, timeout, kick, ban, or route to a human review channel. When the judge isn't confident, it always asks a moderator.
A keyword filter can't read a room.
Matches “kill” → removed. The room loses a normal moment and trust in the bot.
Reads the context, friendly hype between teammates → cleared. A real, targeted threat with the same words would be removed.
Built for real moderators.
Severity you control
Map each of the five levels, sarcasm to severe threat, to the exact actions you want, per server. Combine delete, warn, timeout, kick and ban.
Human review queue
Low-confidence calls and anything you choose go to a review channel with one-click actions. The bot assists; humans decide.
Multilingual & obfuscation-proof
Detection across dozens of languages, resistant to spacing, repeats, leetspeak and Arabic diacritic tricks meant to slip past filters.
Per-server, per-channel
Enable moderation where you want it, set a confidence threshold, pick log and review channels, every setting is scoped to the server.
Kill switch
One toggle halts all automated action instantly. You're never locked out of your own moderation.
Everything logged
Every check, decision and action is recorded with its reason, for transparency, appeals and a clear audit trail.
Pay for judgment, not traffic.
For active, growing servers.
- Everything in Free
- Much higher monthly budget
- Priority judgment
- Extended log retention
- Multiple servers per account
Help shape NestBot, get up to 3 months free.
Read-only, always
Private by design
Up to 3 months free at launch
Questions, answered.
How is this different from MEE6, Dyno or Carl-bot?+
Those rely on keyword lists and simple toxicity scores that read each message in isolation. NestBot sends flagged messages to an AI that reads the surrounding conversation and judges intent, so it clears sarcasm and catches coded, context-dependent harassment that word filters miss.
Does it read every message?+
At launch, a free filter screens every message and the AI judge only sees the small fraction that look suspicious. During the current beta, NestBot runs in analysis mode instead: it collects messages and moderation events from servers that join, pseudonymized, to train the AI. The Security and Data Policy spells out exactly what is stored.
Will it ban people automatically?+
Only if you tell it to. You map each severity level to the actions you want, and you can send anything, or everything, to a human review channel instead. When the AI isn't confident, it always asks a moderator.
What does it cost to run?+
You're billed only for AI judgments, measured in real dollars against your plan's budget. The free tier includes $5 of judgment per month; when it's spent, NestBot falls back to the free gate and routes flags to your moderators.
Which languages does it support?+
The AI judge understands dozens of languages, and the gate's bad-word detection spans many more, including obfuscation tricks like spacing, repeated letters, leetspeak and Arabic diacritics.
Is my server's data used to train models?+
During the beta, yes, that is the whole point of analysis mode: servers that join contribute pseudonymized data to make the AI sharper, and that is what earns the launch reward. You can delete it any time before the beta ends. After launch, messages are processed to make a moderation decision and nothing more; we never sell data.
Connect a server, get
up to 3 months free.
Join the beta by adding NestBot to a server you run. In read-only analysis mode it never bans, kicks, or removes anything, and every member is pseudonymized. Connect a server with 1,500 or more members for 3 months of NestBot free at launch, 1 month for smaller servers.