Hive

Why I built this

Years of investment discussions, market takes, and founder evaluations were buried across hundreds of Slack channels. When someone asked "what did we think about X sector last year?" the answer existed somewhere in the archive — but finding it meant scrolling through months of threads. I wanted to turn that messy conversational history into a structured, searchable knowledge base that the team could actually use.

What it does

Hive archives entire Slack workspaces — messages, files, threads — and runs an AI classification pipeline to synthesize conversations into a searchable knowledge base. Originally built to create a "Founder Bible" from years of Slack history across an investment team.

How it works

• Automated backups — cron-scheduled archiving of all channels, threads, and files
• Message browser — searchable, paginated view of archived conversations
• AI pipeline — classifies threads by topic, then synthesizes related threads into structured articles
• Storage layer — pluggable backend (GCS or S3) with signed URLs and retention policies
• Admin dashboard — manage backup jobs, trigger classification runs, publish articles

The classification step uses a cheaper model for tagging, then a stronger model (Gemini 2.5 Pro via OpenRouter) for synthesis — keeping costs manageable at scale.

Stack

Next.js 15, React 19, Tailwind CSS, shadcn/ui, Supabase PostgreSQL, Google Cloud Storage, OpenRouter, node-cron, Google OAuth