Case study · AI
AI knowledge platform
One conversational layer over 300+ tools and multiple databases — company knowledge you can actually ask questions of.
Problem
The company's knowledge lived everywhere: hundreds of internal tools, multiple databases, documentation scattered across systems. Finding an answer meant knowing which system held it, how to query it, and often who to ask. That works — slowly — until the person who knows is unavailable, and it never scales to AI assistants that could otherwise do the looking.
Approach
I developed, deployed, and project-managed a company-wide AI knowledge platform built MCP-native from the start. The platform wraps more than 300 tools and more than four databases behind a single protocol layer, so chatbots and agents can discover and call any connected capability instead of being hand-wired to each system.
On top of the tool layer sits a centralized knowledge base that consolidates access to company knowledge, connected systems, and internal data intelligence. Agents use it to answer questions, summarize content across sources, and produce reporting over connected data. I managed development and integration across the company's tools throughout — the connective work that makes a platform like this genuinely company-wide rather than a demo.
Architecture
Outcome
Teams across the company query systems they never learned to operate, through agents that discover the right tool for the question. Adding a new system means wrapping it once at the MCP layer — every chatbot, summarizer, and report gains the capability immediately. Company knowledge became consolidated, connected, and conversational.