BuzzWav

Users add subreddits and set scan frequency (hourly to daily). Each scan pulls new posts, scores them for opportunity, and surfaces the top 10 in a ranked feed. Historical data builds a trend line for each subreddit.

Claude Sonnet reads the top posts from a subreddit and generates three content angles per post: what angle to take, why it would resonate, and how to structure the hook. Opportunity score is computed from reply velocity, upvote acceleration, and topic uniqueness.

Every post is embedded via Google text-embedding-004 and stored in Convex. Users can search semantically: 'content ideas about startup failure' returns posts about pivots, shutdowns, and lessons learned even if those exact words never appear.

Momentum scoring tracks reply velocity, upvote rate, and cross-post frequency over a 72-hour window. Posts that spike get a 'Trending' badge with a momentum graph. Users can set alerts for specific thresholds.

Kanban board with AI brief generation: drag an insight into a column, click generate, and get a full content brief with title options, hook variations, key points, and suggested CTA. Briefs can be exported to Notion or copied to clipboard.