Hive Second Brains: Shared Agentic Knowledge/Ontology Infrastructure + Graph RAG System

**Job Description** Our media company has years of Slack conversations, meeting recordings, Zoom transcripts, and Notion documentation. Right now all of that knowledge is trapped in chat scrollback. When leadership needs a decision from six months ago, someone has to search through threads manually. We need someone to design and build an AI system that can: - Ingest Slack messages, Notion pages, Zoom transcripts, and meeting recordings - Extract structured information from unstructured conversations (decisions made, action items, owners, deadlines) - Build a queryable knowledge base the team can access through Slack - Maintain cross-source context so related documents stay connected - Handle more than 200 concurrent users with role-based access The ideal person has experience with: - RAG systems and knowledge retrieval architecture - LLM integration and prompt engineering for extraction quality - Slack API and real-time ingestion pipelines - Vector databases and embedding strategies - Deployment infrastructure This is not a basic ChatGPT integration. We tried that. The hard part is extraction quality. AI tends to flatten information and lose the connections between sources. We need someone who understands how to preserve context across a large interconnected document set.