Quick update on ModPilot AI.
Added Google Gemini integration for moderation summaries. Medium and high risk posts now receive a plain-language explanation describing why content was flagged — for example urgency language, suspicious external links, or promotional patterns.
Core moderation scoring remains fully rule-based. AI is only used to generate readable explanations for moderators. Final moderation decisions always stay human-controlled.
Also continued refining: • moderation queue workflow • action state handling • risk signal tuning • moderator UX polish • live testing environment
Building this publicly as a first-time Devvit developer has been a huge learning experience. Feedback from moderators and developers has been helping shape the project in real time.
App page: https://developers.reddit.com/apps/modpilot-ai GitHub: https://github.com/PrimeWolf-1/modpilot-ai