Inside a Multi-Agent Skill: An AI Research Team of 4 Agents

Free Lesson

Inside a Multi-Agent Skill: An AI Research Team of 4 Agents

Hosted by Sash Mohapatra

132 students

In this video

What you'll learn

What Claude agents and subagents are

An agent runs a task on its own. A subagent is a helper it hands part of the work to.

When one skill is enough, and when you need a team of agents

Tell which jobs a single skill can handle, and which ones call for several agents working together.

How one skill runs a whole team of agents

One skill plans the work, hands pieces to several agents at once, and pulls their results back together.

Meet the four agents: Oracle, Hermes, Athena, Scribe

Oracle plans, Hermes gathers sources, Athena fact-checks, and Scribe writes the report.

Why this topic matters

Running several agents in parallel grew up in coding, where developers split a build across a team. Packaged as a skill, that same orchestration does non-technical knowledge work just as well, like research or a competitive brief. We take apart a working skill with 4 agents, and run it live, so you see how it all fits. All run in Claude code.

You'll learn from

Sash Mohapatra

20 years at Microsoft, engineer to Director of Data & AI Sales, Capital Markets

Sash Mohapatra spent 20 years at Microsoft, moving from customer engineering to Director of AI Sales. He led cloud and ML adoption for Fortune 500 clients. After being laid off in 2023, he used AI to navigate his own career transition and built a system for doing it well.

Now he runs riftlab.ai, where consults for companies on AI adoption - training teams, building agents and helping leaders navigate the AI revolution.

He believes that human ingenuity in the AI revolution can be preserved and scaled through participation. Lean in, learn how to orchestrate AI, be the human in the loop and amplify your leverage and capabilities with AI.

Trained employees from