All posts
AI

How we screen 100% of merchant content with an AI moderator

Every campaign title, every business name, every referral message. Scanned before going live. Here's the keyword + LLM stack that costs $0.0001 per check.

GamifiedDeals·5/12/2026·3 min read

The trust problem

A loyalty marketplace is only as trustworthy as its weakest listing. One scammy 'Free $1,000,000 spin' kills the entire network.

We built a two-stage moderation pipeline that runs on every submission.

Stage 1: keyword pre-filter

A regex-based blocker catches the obvious: slurs, profanity, sexual content, crypto pumps, MLM language, external links. Instant, zero-cost.

Stage 2: Gemini 3 Flash

What the regex misses, an LLM scores 0–100 on policy risk. Verdict 'safe' auto-approves. 'Flag' pushes to admin queue. 'Reject' blocks outright.

Average cost per check: $0.0001. Average latency: 380ms. Coverage: 100% of public-facing strings.

What we won't do

  • We don't moderate private messages between users (privacy).
  • We don't filter by political stance — only policy violations.
  • We don't let merchants moderate their own listings (conflict of interest).
Want this for your shop?

Launch a gamified loyalty program in 30 minutes. Free for 30 days, then $29/mo.

Get started →
Loyalty playbooks, monthly.

1-2 emails a month. Fresh tactics from indie shops that are crushing it. No fluff.