Skip to content
Ryan James
Menu

← Newsletter3 min read

Part 1: Building Engaging PD With Claude Code

A couple of months ago, a few ideas landed at the right time. I’d seen how good Claude is at building out educational curriculum. You can build a whole course with it. Then I learned you can use Claude Code with JavaScript libraries to generate video content, so you pick the libraries and it builds training videos for you. I’d already used different LLMs to build small web apps. So I thought, let’s put all of these pieces together.

I’m the Director of Technology at a K-12 independent school, and we have a lot of technology. A phone system, email, copiers and scanners, a PA system, AI tools, classroom tools, virtual meeting tools. It takes a lot of expertise to use all of it well. On top of that, many of our employees need professional development hours to keep their teaching certifications. Teachers want PD, but they don’t want to stay after school, and they don’t want to give up classroom time. I wanted to build something that solves that.

Getting the content in

Last year I built a NotebookLM and dumped in years of training slides, handouts, and how-tos, the things we’d been emailing out whenever we rolled out a new phone system or someone had a question. This time I went to Claude first and asked it to write the prompts for NotebookLM. NotebookLM then generated the professional development content, which was really built from material created by many people on my team, including me. I took that content and gave it to Claude Code, almost like getting two AIs to talk to each other. Claude Code used it to build the framework.

I organized it into living buckets. Google Workspace is one bucket, and inside it are modules on each of its features. I hadn’t used Workspace in years before I came to my school, so I know what it’s like to figure it out as you go. Most people can get by. But are they using all the features? Probably not. That’s what I want people to learn.

Locking it down and tracking it

The content sits behind Google single sign-on, which I set up using my admin privileges. When someone finishes training, it’s saved to a database. They can print a PDF certificate straight from the app for their PD documentation, and from the admin side I can run a report on who’s done what. That lets me work with supervisors, whether that means setting an hour target for their teams or assigning specific modules. There’s also a required set of about seven modules, roughly three hours total, almost like a boot camp that unlocks the rest. I added badges and a leaderboard on top of that. Finish every module in cybersecurity, for example, and you earn that badge.

My first version was boring

The first module Claude generated was a PDF walkthrough, an optional video, and a quiz. I’ve taught classes and I’ve taken a lot of online training, and I looked at it and thought, nobody is going to want to do this. You’d read it or play the video, take the quiz, and learn nothing.

What fixed it was what I knew as a teacher: we learn by being curious and by doing. I started prompting Claude to build modules that spark curiosity and make people take the actions, and to require that work before the quiz unlocks. Now, to learn how to set a vacation responder, you go through a mock email setup and set one up. You’re doing the steps, not reading about them. I didn’t want a tool where someone presses play, goes to get coffee or make copies, and comes back to take the quiz. Realizing I could bring good teaching into this was my aha moment.

The honest part

It’s a lot of work, with a lot of back and forth with Claude, and I’ve only touched it off and on over a couple of months. There are weeks I don’t get to it at all. It’s about the same effort as prepping a week of PD courses, which is substantial. But a week of PD gets delivered once. This will keep paying off because people can go at their own pace, stop, and ask questions, without setting up a one-on-one with me or waiting for a department meeting.

I’m almost done building. My plan is to finish the courses by the new year, pilot in the spring, and roll out to all employees next summer. In part two I’ll write about finishing the build and starting the pilot. In part three I’ll look back on the rollout and what I’d do differently. I’m curious what people will think of it.

Keep coding!
Ryan

This post first went out on Substack. Join the conversation there or subscribe to get the next one by email.