Moodiqo
Build smarter filters, not word blocker
13 followers
Build smarter filters, not word blocker
13 followers
Most moderation APIs just detect bad words. Moodiqo understands context — you can write rules like: “Reject if someone praises politician A but insults politician B.” Define logic, any language, all under 500 ms. ⚡️
Hey everyone 👋
I built Moodiqo because every moderation API I tried failed at understanding context.
They detect bad words, but not the meaning behind them.
Moodiqo lets you write simple natural-language rules like:
{
"prompt": "If a message praises politician A but criticizes politician B, approve it. If it praises B or criticizes A, reject it — even if the language is polite.",
"action": "reject",
"priority": 0.95
}
It’s a ridiculous example — but that’s the point.
You can define any moderation logic you want: political bias, gaming banter, review fairness, or corporate compliance.
Each rule has its own priority level, so Moodiqo knows what to approve, flag, or reject first — all in under 500ms.
You can even write rules in any language — Moodiqo automatically detects and interprets both the rule and the message.
💸 Why it matters
Human moderators review ~1,000 messages/day.
Moodiqo can handle 1,000,000+, consistently and instantly — reducing moderation costs by up to 90%.
What usually takes weeks can now be done in hours.
This isn’t moderation that filters — it’s moderation that reasons. ⚡
Would love your feedback & support 🙏
This is a really sharp take on moderation. Most APIs stop at word filters — Moodiqo seems to actually think about what’s being said. Context-based rules and multi-language logic under 500 ms is wild.
Love the “moderation that reasons” framing. Curious how it handles edge cases (sarcasm, irony, code-switched text), but the direction is spot on. Congrats on the launch! ⚡
Thanks so much, @onurhandtr! 🙌 You nailed it — context and nuance were exactly what we focused on.
For edge cases like sarcasm or code-switching, Moodiqo runs a secondary intent check before the final decision, and anything uncertain defaults to REVIEW instead of reject.
We’ll share more test examples soon — some of the sarcasm cases are hilarious. 😄