A moderation bot that covers the hours your mods can't.
Your server doesn't go quiet when your team logs off — it gets louder. Wardn watches the whole 24 hours, handles the obvious calls the way your mods would handle them, and hands anything borderline to a human with the context attached.
We onboard five servers per batch. Each one is quoted on its own message volume and setup.
10 hrs
a day with zero human eyes on the server
41%
of the day's traffic lands in those hours
1 tap
to reverse anything Wardn did while you slept
Every server outgrows its first moderation bot the same way.
Not because the bot broke. Because the server got big enough that the interesting cases stopped being the ones a rule can describe.
Stage one: the wordlist works
Two thousand members, one filter, a handful of banned words. Anything that slips through gets caught by a human in ten minutes.
Stage two: the arms race
Someone writes 'fr33 n1tro'. You add it. Next week it's spaced out with emoji between the letters. Your filter becomes a 400-line file nobody wants to touch.
Stage three: judgment, not matching
The hard calls are now about intent. Is this artist self-promoting or contributing? Is this a joke between friends or harassment? No rule answers that. A trained reader does.
What each kind of bot can and can't do.
We're not going to pretend the classics are useless — they're excellent at the jobs they were designed for. They were just designed before anyone needed a bot to look at a picture.
| The job a moderation bot is actually hired for | Wordlist bots | Native AutoMod | Dev-first AI tools | Wardn |
|---|---|---|---|---|
Catches the same spam after one character changes f r e e n i t r o, sc4m, Cyrillic lookalikes | ✕ | ✕ | ~ | ✓ |
Reads what's inside an uploaded image text baked into banners, QR codes, NSFW uploads | ✕ | ✕ | ~ | ✓ |
Follows a link to where it actually lands shorteners, redirect chains, look-alike domains | ✕ | ✕ | ~ | ✓ |
Checks usernames and display names offensive handles, names used as free advertising | ~ | ✕ | ~ | ✓ |
Explains every action in a sentence a mod can argue with not a confidence score, an actual reason | ✕ | ✕ | ✕ | ✓ |
Gets better when a mod overturns it the override is the training signal | ✕ | ✕ | ✕ | ✓ |
Set up without anyone touching config files your mod team, not your one technical friend | ~ | ✓ | ✕ | ✓ |
Levels, welcome messages, reaction roles, tickets — your existing bot still does those better than we ever will. Wardn takes the moderation layer only.
Where a wordlist genuinely fits — invite links, hard-banned slurs — keep it. Wardn handles everything that needs a second look, and everything AutoMod structurally can't see.
Other tools ask you to describe your rules. Wardn reads how your team has already been moderating and learns the rules you never wrote down.
Five servers per batch, quoted one at a time.
Setting this up properly means reading your history, sitting in your channels, and tuning against your team's calls. That doesn't scale to a public invite link — so it doesn't have one.
Why the batch is five
Each server gets its own tuning pass with a real person on our side. Five is what we can do without any of them getting the leftovers.
Why there's no price on this page
A 6k-member hobby server and a 300k-member game community are different amounts of work. There's a setup fee and an ongoing fee, and both are quoted after we've looked at your volume and how you want it wired up.
Questions we get on the first call.
Do we have to remove our current bot?
No. Most teams keep it for utilities and hand only the moderation decisions to Wardn. We'll tell you which of its filters to switch off so you don't get double actions.
What permissions does it need?
None on day one. It reads for about a week, you review what it would have done in a sandbox, and only then do you grant the actions you're comfortable with.
What if our mods disagree with a call?
They overturn it in one tap, and that overturn is the training signal. The point isn't a bot that's always right — it's one that's wrong in the same way only once.
Is this only for huge servers?
It's built for teams where volume has outgrown people. That's usually five figures of members, but a fast 4k server can be worse than a sleepy 60k one.
Tell us about your server.
The more you tell us about volume and what's actually hurting, the more useful the first call is. We'll come back with what we'd do and what it costs.
Go deeper on the part that's hurting
Every server's worst problem is a different one.
AI Discord moderation bot →
The training loop: every override your mods make becomes a rule you never write.
Discord AutoMod alternative →
Why wordlists lose to one swapped character, and what reads intent instead.
Discord scam & phishing protection →
Unmasking redirect chains, fake giveaways and QR drainers before members click.
Discord content moderation →
NSFW, gore and slurs handled the same way at 3am as at 3pm.
Discord self-promotion bot →
Telling the artist sharing work apart from the person farming your members.
Discord raid protection →
Lockdown that doesn't punish every real member who happened to join today.