Robyn AndersonCEO, Jam Ai · AI Implementation & Business Development Leader

Part 2/5 How to make an AI Agent with Jam

13 Jun 2026 · 1 min

Watch on YouTube

1:04

In this second installment of the "How to Build an AI Agent with Jam" series, Robyn Anderson walks viewers through the workflow of converting the handwritten agent design notes created in part 1 into structured markdown files using Claude Cowork. She explains how to upload photos of the notes, prompts Claude to extract each agent’s role, required connectors, content, and the necessary MCP (Minimum Capabilities Package), and then generates an overview for every agent. The segment emphasizes that these markdown overviews will later be used to build the actual agent files in Obsidian (covered in part 3). Robyn closes with a reminder to follow and like the channel for algorithmic boost.

Chapters

Welcome back to part 2 of how to build an AI agent. You might be wondering why I'm holding a pen and paper. Well, if you are, go back and watch part 1 because yesterday you wrote down exactly what AI agents you wanted to build. And today we're going to throw pictures of our chicken scratch notes into Claude Cowork and actually have it build an MD file about every agent that you described. So step 2 is to upload pictures of all of your agents that you wrote about. Remember, you wrote their role, their connectors needed, and the content needed. And you're just going to say, I want to create an agent system for my business. I wrote down the agents by things I do. Here are pictures of my notes. Help me organize what my agent team would look like. I need you to figure out what MCP is needed and what content is needed for each agent, because obviously it's going to add to what we thought it might need. From there, it'll actually create, uh, an overview for each agent and what they need. From there, we're going to build out our files in Obsidian, but that's for part 3. So join me tomorrow, hit follow and like so the algorithm will do its thing.