SCOUT: Institutional Knowledge Assistant

New hires lose 4–5 hours per day for their first two weeks hunting through scattered Slack threads, outdated Confluence pages, and shadow docs just to find basic answers like 'What's our deploy process?', 'Who owns the payment service?', or 'Why was this built this way?'

A unified RAG assistant that ingests an entire institutional knowledge stack (Slack history, Notion docs, GitHub wikis, Jira tickets, Figma comments, email threads) into a vector store, contextualizes by role and project, and answers natural-language questions with citations and the people who wrote the source docs.

• Connect Slack, Notion, GitHub, and Jira APIs (auto-syncs nightly)
• Ask a question in Slack DM or the dashboard, e.g. 'How do I run the auth service locally?'
• Multi-source retriever queries the vector DB and ranks by recency and user role
• Receive synthesized steps with links to source docs and tagged subject-matter experts
• Thumbs-up/down feedback improves retrieval ranking over time

• Multi-source retriever — across Slack, Notion, GitHub, Jira, and more
• Contextual ranking — by user role (Backend / DevOps / Sales) and project relevance
• Reference generator — answers cite specific source docs and authors

• RAG orchestration: LangChain / LlamaIndex
• Vector storage: Pinecone / Weaviate (multi-tenant)
• Data ingestion: Slack Bolt API, Notion API, GitHub API
• Frontend: Next.js dashboard
• Vibe coding: Claude Cowork