Gettit — AI Moderation Assistant for Reddit

An AI-powered Devvit moderation assistant that helps Reddit moderators analyse discussions, detect harmful content, and moderate communities faster.

Comment

Inspiration

Reddit moderators handle massive amounts of discussions every day, and manually reviewing threads for toxicity, spam, harassment, or suspicious activity can become overwhelming. I wanted to build a lightweight AI-powered moderation assistant that helps moderators quickly understand discussions and identify potential risks without replacing human judgment.

This project was inspired by the idea of making moderation workflows faster, smarter, and more efficient using AI.

What it does

Gettit is an AI-powered Reddit moderation assistant built with Devvit and Gemini AI.

The tool allows moderators to analyze Reddit threads directly from the moderation menu and instantly receive:

  • Thread summaries
  • Discussion topics
  • Risk assessment
  • Concerns detected in the conversation
  • Flagged terms
  • Moderation recommendations

The goal is to help moderators review discussions more efficiently and identify problematic content faster.

How I built it

I built the project using:

  • Devvit
  • TypeScript
  • Hono
  • Gemini 2.5 Flash API
  • Reddit Dev Platform APIs

The app works by fetching Reddit post comments, sending the discussion to Gemini AI, parsing the AI-generated moderation report, and displaying the results inside a clean moderation popup UI.

To improve performance and reduce API usage, I also implemented in-memory caching for repeated thread analysis.

Challenges I ran into

One of the biggest challenges was handling AI response formatting reliably.

Initially, I tried forcing strict JSON responses from Gemini, but the output became unstable during testing. I later redesigned the system to use structured section-based parsing instead of strict JSON parsing, which made the moderation analysis significantly more reliable.

Another challenge was working within Devvit form UI limitations while keeping the moderation popup readable and compact.

I also had to carefully optimize prompts and token usage to avoid exhausting Gemini API quota during testing.

What I learned

During this project, I learned:

  • How to build Reddit moderator tools using Devvit
  • How to integrate Gemini AI into production-like workflows
  • Prompt engineering for stable AI outputs
  • Handling AI response parsing safely
  • Designing moderation-focused user experiences
  • Managing API limits and caching strategies

What's next for Gettit

Future improvements could include:

  • Real-time moderation monitoring
  • Automatic moderation actions
  • Advanced toxicity classification
  • Moderator dashboards and analytics
  • Multi-language moderation support
  • AI-powered spam detection pipelines

Gettit is designed as a foundation for smarter AI-assisted Reddit moderation tools.

Built With

Share this project:

Updates