The situation
Knowledge at the 200-person company was fragmented across Confluence, Notion, and scattered internal docs. Employees wasted time hunting for information, and AI tools had no unified way to access it.
What we did
We built an MCP server in TypeScript that unifies Confluence, Notion, and internal documentation into a single AI-accessible context, indexed with Elasticsearch and connected through the Confluence and Notion APIs.
The outcome
Information search is 80% faster, answers on internal documentation are 95% accurate, and employees save around 2 hours a day.