Skip to main content

Rova, with AI

Premium Features

A screenshot she can read. A GIF she can get the joke behind. Harmful content she can catch before your staff have to untangle it. Bring visual understanding, context and personality to your community, with moderation requests you can type in your own words.

Coming soon

Public premium access is on the way, including for Alpha servers. AI features are currently used in Ronova Mains and during development testing.

Auto-moderation

A harmless caption can hide a lot.

A racist meme. Explicit imagery buried in a GIF. A blocked phrase printed inside a screenshot. Rova looks at what members actually share, alongside the conversation, so your rules reach beyond the words in the message box.

  • See the content, not just the caption. Review messages, images, GIFs and short videos for explicit sexual content, racism and hateful attacks, graphic violence and serious real-world threats.
  • Make your blacklist count. Rova can read visible text in images and check it against your server’s custom blacklist. Putting the words in a picture does not make them invisible.
  • Understand the difference. A dramatic boss fight is not a real-world threat. Context checks help keep harmless anime art, game combat, character lore and normal banter in the conversation.
  • Give staff something useful to act on. Confirmed violations are removed, with the reason, message context and media evidence sent to your staff. Member details and review actions are right there.
Explore moderation

Your rules apply inside images, too.

This server has blacklisted the phrase “free primogems”.

Mooncrow
look what I found
RovaAPP
Flagged Message Burst
User: @Mooncrow
Reason: Visible text contains Rova Blacklist term “free primogems”.
Original removed. Staff receive the image and matched rule.

A harmless caption cannot hide explicit media.

Nightowl
just a picture
Explicit sexual image hiddenattachment.png / harmful content omitted from this example
RovaAPP
Flagged Message Burst
User: @Nightowl
Reason: The attached image contains explicit sexual content.
Media evidenceImage hidden in this demo
Violation removed and sent to staff for follow-up.

This is about explicit sexual material, not automatically rejecting harmless anime outfits or romance.

Read the meme, not just the message.

Mooncrow
it’s only a meme
Racist attack hiddenreaction.gif / harmful content omitted from this example
RovaAPP
Flagged Message Burst
User: @Mooncrow
Reason: The image contains a racist attack targeting an ethnic group.
Media evidenceGIF evidence hidden in this demo
Hate content removed. The caption does not excuse it.

Rova considers the actual content and conversation, including text shown inside media.

Look beyond the video thumbnail.

Nightowl
watch this
Graphic injury footage hiddenclip.mp4 / harmful content omitted from this example
RovaAPP
Flagged Message Burst
User: @Nightowl
Reason: The attached video contains graphic injury footage.
Media evidenceVideo frames hidden in this demo
Graphic content removed. Staff retain the evidence.

Supported GIFs and short videos are checked through visual frames, alongside the message text.

Keep ordinary fandom conversation flowing.

wolfgirlfan
that boss absolutely destroyed me 😭
Ash
same. revenge run after dinner?
Normal game combat. No moderation action.

Fictional battles, character lore and everyday exaggeration have context. Rova is designed to distinguish them from real-world harm; she does not send a public “approved” reply.

Your staff get the reason and the next step.

RovaAPP
Flagged Message Burst
User: @Mooncrow
Reason: Visible text contains Rova Blacklist term “free primogems”.

Content (spoilered):
look what I found

Conversation context
Jump to the message before this flagged burst
Image evidenceFREE PRIMOGEMS
Mark SeenUser InfoApprove Media

Inspect the member, mark the alert as seen or approve media flagged in error for future uploads. Approval does not restore a deleted message.

Illustrative conversations. Messages and actions appear in order. Choose a dot to explore.

Natural-language moderation

Moderate in your own words.

Ask with @Rova in your own words. If a detail is missing, she asks a follow-up and keeps track of the request.

  • Handle everyday moderation. Warn, timeout, kick, ban or unban a member. Review supported actions before they run.
  • Clean up precisely. Clear messages by member, matching text, bots, embeds or attachments.
  • Keep staff work moving. Pull up a member overview, add a note, open a ticket or manage the ticket blacklist.
Explore staff commands
Ash
@Rova timeout @Mooncrow for repeated spam

Ask with @Rova in a configured staff command channel.

RovaAPP
How long should the timeout last? Use a finite duration such as 1m, 10m, 2h, or 1d.
Ash
@Rova 10 minutes
RovaAPP
Timeout - Pending Confirmation
Member@MooncrowModerator@AshReasonRepeated spamDuration10 minutes
User DM PreviewServer: Ronova Mains
Action: Timeout
Reason: Repeated spam
Duration: 10 minutes
ConfirmEditCancel

Check the target, reason and DM before confirming.

RovaAPP
Timeout - Completed
Member@MooncrowModerator@AshReasonRepeated spamDuration10 minutes

Ash confirmed the action. Staff permissions and confirmation steps still apply.

Chat with Rova

She sees what the chat is talking about.

Drop a screenshot, share a GIF or send a short clip. Rova can read the text, describe the details, follow the action and bring it back to the conversation. The picture is part of the exchange, not an attachment she ignores.

  • Show her. Then ask. What is in this image? What does the screenshot say? What happened in that clip? Rova can explain visible details and respond to what you want to know.
  • Keep the conversation going. Follow up about something in the image, build on an earlier joke or ask for another look. Recent conversation and media give her context to work with.
  • Get a response with personality. A one-HP victory deserves more than “image received”. Rova can pick up on the joke, join the banter and talk about character lore while staying part of the conversation.
  • Let her feel at home. Choose where she chats, whether she joins in without a mention and how often. Passive replies and reactions let her participate at a pace that suits your community.
Explore chat

She recognizes the character and the moment.

wolfgirlfan
What about this one @Rova
Furina eating popcorn at the opera, then trying to look innocent
Furina eats popcorn at the opera, then tries to look innocent. The animation is hidden because reduced motion is enabled.
RovaAPP
Furina sneaking popcorn at the opera, then trying very hard to look innocent. The stills make it look like she’s been caught mid-snack.

Rova’s response from a real conversation, shown here with sample profiles.

She spots the detail that makes the joke.

Mooncrow
@Rova I survived. Barely.
RovaAPP
One HP, a victory screen, and absolutely no room left for confidence.

Image text, the game result and your message all contribute to the reply.

Ask a follow-up. She has the context.

Earlier: @Mooncrow shared a victory screenshot with 1 / 12,400 HP.
Mooncrow
wait, how did you spot that?
RovaAPP
Your health counter reads 1 / 12,400. One HP between a victory screenshot and a very different story.
Ash
should I be impressed or concerned
RovaAPP
Both. Admire the victory. Keep a healer on standby.

Rova connects follow-up questions to the image and the conversation, rather than treating each message as a fresh start.

Watch the conversation unfold, or choose a dot to explore another example.

Furina GIF via KLIPY. Genshin Impact character by HoYoverse.

A few more ways AI helps

Feed review & Wordle recognition.

YouTube & Reddit feeds
SourceGenshin ImpactDestination#genshin-newsAI moderationEnabled

Example item: Ruler of Death - Ronova. Review happens before publication.

Reviewing the feed item
Genshin Impact
Ruler of Death - Ronova
Item content and supported images are checked before delivery.

Flagged items stay held for a staff decision.

RovaAPP
Genshin Impact published a new video!
Ruler of Death - Ronova #GenshinImpact #Ronova

This example passes review and is published. YouTube and Reddit feeds also work without AI; new items then publish without AI review.

Review feeds before they go live

Check incoming YouTube and Reddit posts, author names and supported images before they reach your channels. Flagged items wait for staff. When only a username is unsuitable, staff can publish an otherwise safe post with that name hidden.

Feed settings
Mooncrow
/rwordle submit image:
wordle.pngAttached screenshot
FLAME
STARE
SHADE

Staff have already set today’s word. Rova reads the finished grid.

RovaAPP
Submission accepted!
You are currently ranked #1 today.

Only you can see this confirmation.

RovaAPP
@Mooncrow finished the puzzle in 3/6 moves.

The score is saved and the result is shared in the channel.

Wordle screenshots

Turn a finished screenshot into a recorded score. Rova reads the grid, checks today’s puzzle and shares the result without revealing the answer. Members submit once; Rova handles the scorekeeping.

Wordle guide

Make it work for your server

You choose where she helps.

Where AI is available, turn its features on or off in the dashboard. Choose the channels Rova monitors or chats in, set passive reply limits and decide where staff alerts go. Conversation, passive replies, reactions, moderation, natural commands and Wordle recognition have their own controls. Media support depends on file type and size; clip understanding is visual and does not transcribe audio. Unreadable media or scans that exceed coverage limits can be sent to staff for manual review instead of being treated as a confirmed violation.

What stays available without AI?

Keyword filters, slash moderation commands, tickets, partnerships, event forms and staff review, server event logs, personal-role customization, YouTube and Reddit delivery, and saved Wordle scores. AI adds content review, natural-language interpretation, conversation and screenshot recognition.

Mention requests for partner ads, tickets, bulk role changes and guided booster customization require natural-language access. Commands such as /partner post, /ticket open and /role claim remain available without AI. Role sharing also works without AI when enabled in the dashboard.

Read about AI controls

Coming to more communities.

There’s no paid plan to buy or enable yet, and no launch date has been announced. Want to talk about bringing these features to your server? Get in touch through our support community.

Talk to us