Pre-launch

Discord moderation

Moderation you can show your work on.

Every moderation bot can remove a message. The question that comes up three weeks later is why it did, who agreed, and what happened when somebody complained.

The part that goes wrong

A log is not a record.

Most moderation tooling produces a list of things that happened. An action, an account, a timestamp. It reads like evidence right up until somebody disputes one, and then you notice it contains no reason, no evidence and nobody's name against it.

The gap only shows up under pressure, which is the worst time to find out your history is a receipt roll rather than a case file.

Why is a Discord audit log not enough for moderation?

Because it answers what and when, and moderation disputes are about why and who. Discord's audit log records that an action occurred and which integration or account performed it. It was designed to tell you what changed in a server, not to hold the reasoning behind a judgement about a person.

For most of what a server does that is completely fine. It stops being fine at the exact moment it matters: a member says they were removed unfairly, a moderator says they were following policy, and the only artefact anybody can produce is a line saying a timeout was applied. Both of them are now arguing from memory.

What is different here

Every action carries its reason.

A moderation action is not an event in a stream. It is a decision, and a decision has an owner, a basis and a way to be challenged. That is the shape SwarmUp stores it in.

What does SwarmUp record for a Discord moderation action?

These fields, on every action, whether a rule fired it or a person did:

  • the community
  • the sender
  • what detected it
  • a confidence value
  • the evidence
  • the action taken
  • who owned it

Appeals and undo are retained, and the reversal is recorded too. So the history shows not only what your moderation did but where it was overruled, which is the only version of a record that is worth anything in an argument. A history with no corrections in it is not a sign of a system that never erred.

Where judgment goes

Held, not guessed.

The interesting failure in automated moderation is not the thing it gets wrong. It is the thing it should never have decided alone. Patch applies the rules you wrote, holds what needs judgment, and hands the high-risk calls to a person.

What happens when a moderation decision needs a human?

It goes on a queue and waits. Incidents and escalation run on a real queue rather than as a notification somebody may or may not read, so a held decision has a state and an owner instead of sitting in a channel scrolling out of view. Deterministic low-risk rules keep acting automatically in the meantime.

Drawing that line is the actual configuration work, and it is a policy decision rather than a technical one. Set it too low and your moderators are approving obvious spam by hand at two in the morning. Set it too high and the system is making permanent decisions about people on its own, which is the thing it is specifically not allowed to do here.

And the same rulebook runs your Telegram

One policy, two platforms.

Discord bots are Discord only. Telegram bots are Telegram only. Most projects run both and hold the connection between them in a moderator's head, which works until that person is asleep.

Can one moderation policy cover Discord and Telegram?

Yes. Both communities are operated from one workspace and run on one Community Blueprint, which stores the rules, tone, knowledge and goals you explicitly approve. Each platform and each channel still keeps its own permissions, so one policy does not mean one undifferentiated set of behaviours everywhere.

What it does mean is that an account you acted on in one place is not met as a stranger in the other, and that changing your mind about a rule is one edit rather than two edits and a note to remember the second one.

What it costs

Safety is not the upsell.

The standard shape of this market is that protection is the hook and everything that makes protection accountable is the paid tier: history behind a plan, appeals behind a plan, a second server behind a per-server price.

Is Discord moderation a paid feature here?

No. Core protection is not a premium feature, community growth is not metered, and only AI generation you ask for uses credits. These are unmetered at every volume, on every plan:

  • Live moderation
  • Scam and impersonation detection
  • Raid and anti-nuke protection
  • The incident queue
  • Appeals
  • Audit history
  • Sentiment and analytics
  • Setup

The reason to build it this way is not generosity. A bill that scales with moderation volume charges a community most on its worst day, and a product that does that has quietly given its customers a reason to moderate less.

Where the automation stops

What the AI is not allowed to do.

The reason to trust a moderation product is not how clever it is. It is what it is prevented from doing without you.

  • AI cannot ban a member on its own.
  • AI cannot apply a permanent sanction on its own.
  • AI cannot publish a change to your policy.
  • Deterministic low-risk rules act automatically. Everything above that line waits for a person.
  • Every action is reversible, and the reversal is recorded too.
  • Provider identities are stored as one-way pseudonyms, not as raw platform accounts.

Straight answers

The questions moderators actually ask.

What is a Discord moderation bot?

A Discord moderation bot enforces a server's rules automatically: removing content, timing out or removing members, and flagging what needs a human. SwarmUp does that under a policy your team writes, and keeps a reviewable record of every action, including the ones that were later undone.

Can a moderation bot ban members automatically?

Most can. SwarmUp will not. AI cannot ban a member on its own and cannot apply a permanent sanction on its own. Deterministic low-risk rules act automatically, and everything above that line waits for a person, because a permanent decision is the one you cannot take back cheaply.

What happens when the bot gets a moderation decision wrong?

You reverse it, and the reversal is recorded too. That second half is the part most tooling skips. A system that quietly corrects itself leaves you unable to show that you corrected anything, which puts you in the same position as never having corrected it at all.

Can moderators see why an action was taken?

Yes. Every action carries the community, the sender, what detected it, a confidence value, the evidence, the action taken and who owned it. That is what a moderator opens when a member appeals, rather than reconstructing the reasoning from a timestamp and a rule number.

Does it work across more than one Discord server?

Yes. Servers, groups and seats are unlimited on every plan, including the free one. Plans are separated by how many workspaces you run, not by how many servers or members you have, so a second community does not change the price of moderating the first.

Is Discord moderation available today?

Moderation runs today and is the core of the product, but the workspace is not open to the public yet and nothing is charged before it opens. Engagement and rewards are still in testing. Where each part stands is stated on the page rather than implied.

Where we are today

SwarmUp is not open to the public yet. Moderation is the core of the product and it runs today. Engagement and rewards are still in testing. Plans and prices are final, and nothing is charged before the workspace opens.

One email, when it opens

We will not waste the one email.

We are not naming a launch date until we can hold it. What you get for an address is a single message the day the workspace opens, and the same plain account of where the product stands on that day.