One AI Chat to Rule Them All: Connecting Enterprise Tools with MCP
Byte Size (BEGINNER level)
Room C
Tired of juggling Jira, Confluence, GitLab, and Slack/Mattermost? What if one AI could search tickets, find docs, check PRs, and summarize discussions, while keeping data on-premise and avoiding subscription costs?
I'll show how to build this using the Model Context Protocol (MCP), an open standard for AI-tool integration, now under the Linux Foundation and adopted by OpenAI, Google, and Microsoft.
What the audience will see:
MCP Introduction : What is MCP, how old is it, who's driving it forward (Anthropic, growing community), and why it matters - the "USB-C for AI tools."
Architecture and Implemetation : How MCP clients, servers, and the protocol work together. The local stack: Ollama + Open WebUI + mcpo proxy connecting to MCP servers for Atlassian, GitLab, and Obsidian.
Demo : Real queries across systems: Searching tickets, finding docs, checking repos
Key Takeaways :
I'll show how to build this using the Model Context Protocol (MCP), an open standard for AI-tool integration, now under the Linux Foundation and adopted by OpenAI, Google, and Microsoft.
What the audience will see:
MCP Introduction : What is MCP, how old is it, who's driving it forward (Anthropic, growing community), and why it matters - the "USB-C for AI tools."
Architecture and Implemetation : How MCP clients, servers, and the protocol work together. The local stack: Ollama + Open WebUI + mcpo proxy connecting to MCP servers for Atlassian, GitLab, and Obsidian.
Demo : Real queries across systems: Searching tickets, finding docs, checking repos
Key Takeaways :
- Basic understanding of MCP protocol
- See a real POC connecting to enterprise tools
- Blueprint of the architecture to experiment with