Weekly Newsletter

Practical AI Strategies

A Brand New Website for Teaching AI Ethics

Plus: What is MCP, and why should you care?

Hi everyone,

For the past couple of weeks, I've been experimenting a lot with Claude Cowork and Claude Code. If you're not familiar, these are applications designed to make working with the Claude AI model more useful in a variety of contexts. Cowork is a feature in the Claude Desktop app (formerly MacOS only, but they just released a Windows version). It lets you work in a folder on your device, and can edit and manipulate files directly.

Claude Cowork is the "user friendly" version of Claude Code, which runs in the terminal (command line) and can similarly interact directly with your machine. These are potentially risky, but very powerful applications.

To test Claude Code, one of my first ideas was to bring to life a project I've had on the back burner for a long time: turning the Teaching AI Ethics resources into a standalone website.

Teaching AI Ethics: The Website

Building and maintaining a website is no easy task. Even if you use "low code/no code" tools like Wordpress or Squarespace, there's a lot involved. But Claude Code is excellent at writing code (of course), and that includes websites. I'm going to write a longer post about this next week, which will include a discussion of the ethics of using GenAI to build a website about the ethics of AI...

TL;DR: I took my two free Teaching AI Ethics eBooks, put them into a folder, added some instructions on the architecture of the website, and hit send. Claude Code worked semi-autonomously for an hour, building and then deploying (via GitHub) the entire website.

I then made some manual edits, and ran a few checks for security, usability, and so on. I also had Claude Code do a check for WCAG accessibility compliance. In total, I had the website up and running in less than half a day.

I launched the new website via my blog this week, alongside the new free eBook which you should have received a copy of (check the weekend's emails!)

Read about the Teaching AI Ethics launch on the blog

What is Model Context Protocol?

Another important aspect of Claude - and increasingly all GenAI platforms - is something called the Model Context Protocol (MCP). This is a web standard, originally published and then open sourced by Anthropic, Claude's parent company. It is essentially a bridge between AI apps like Claude and ChatGPT, and all the other digital widgets we use every day.

You've probably "connected" apps before, maybe using your Google account to Single Sign On to another platform, or allowing Facebook to read data from another app. MCP offers a way to invite GenAI into that party. It means you can potentially connect any of your digital systems to a model like Claude. In this week's post, I explain how (and why) I connected my todo list app, my client database, and my Wordpress blog to Claude by building my own MCP servers.

It's a bit technical, but I really want you to understand the underlying technology because it is already becoming one of the most important features of GenAI.

Read about MCP on the blog

Hope that you enjoy this week's articles! Remember, I read and respond to every email from this list so feel free to get in touch by hitting reply to this email!

Cheers,

Leon

PS: A couple of weeks ago I asked for feedback on a new idea. Based on hundreds of responses, I landed on a "library" of curated resources. It has different shelves for research, AI updates, AI news, and practical how-to articles and videos.

I keep the Practical AI Library curated and up to date - anything you see in the library is something I've read, appreciated, and maybe referenced myself.

Before the public launch in March, this mailing list and my online course members can access the first month for just $4 AUD. Almost 300 educators have already signed up.

​You can join them here

Here's an example of the 'Academic Research' shelf:

---

---