Autonomous Operations with MCP and AI Agents
This is the cross-functional path. Participants build three structured agents — request intake, knowledge-grounded Q&A, multimodal meeting summary — and then a fourth completely freehand agent that they design from scratch. The freehand lab is the hackathon-vibe finisher the customer asked for, and is especially powerful here: cross-functional learners often have the clearest picture of which everyday task they want an agent to take off their plate.
.svg){kind=small}
About The Lab
Prerequisites
Audiences
Lab Architecture
The Project architecture integrates Claude Code with a multi-model layer (Claude Haiku, Pixtral-large for vision, faster-whisper for audio) to create a comprehensive environment for autonomous operations with MCP and AI agents, which includes four specialized labs Internal Request Intake, Knowledge-Grounded Q&A, Multimodal Summary, and Cross-Team Handoff Orchestrator that allow the user to test end-to-end operational use cases, from routing requests and answering questions with cited sources to summarizing multimodal inputs and producing approved cross-team handoffs.
Why this Lab ?
This is the cross-functional path. Participants build three structured agents -- request intake, knowledge-grounded Q&A, multimodal meeting summary -- and then a fourth completely freehand agent that they design from scratch. The freehand lab is the hackathon-vibe finisher the customer asked for, and is especially powerful here: cross-functional learners often have the clearest picture of which everyday task they want an agent to take off their plate.
Lab Objectives
- Build three cooperating cross-functional agents following structured guidance, then a fourth freehand agent.
- Use real model routing for triage, knowledge-grounded answers, and multimodal summaries.
- Combine static documents with live mock MCP data covering tickets, calendars, and ownership.
- Practice human approval gates so the agent never takes a write action without explicit confirmation.
- Finish the project with one freehand agent that handles a real Monday-morning chore the learner brought.
