TODAY’S THOUGHTS ☠️

Hey friend 👋,

I’ve noticed an interesting shift between casual users of AI, aka those who mostly use chat, and power users of AI, who are using it for completing tasks and light-mid automation.

And that is the use of context and skills.

When I say skills, I don’t mean the “skills” we cover in L&D but teaching skills to AI. It’s becoming an emerging requirement of meaningful use of AI. Without understanding how to feed AI your context and deconstruct skills that AI can use, you’ll be stuck firmly in the ‘average’ user level.

I don’t believe that’s good for you, your skills or your career.

If we want L&D to be taken seriously as “The AI enablement” function in the workplace, you need to add this to your talent stack.

Today, we're exploring why you should learn to build skill.md files to unlock AI’s full power.

Get your tea or beverage of choice ready 🍵.

We've got lots to discuss!

P.S. Your app might clip this edition due to size. If so, read the full edition in all its glory in your browser.

IN THIS DROP 📔

🙋‍♀️ Want to reach over 5,000 L&D pros? Become a Newsletter sponsor in 2026

THE BIG THOUGHT 👀

Things just keep changing so fast, right?

We’ve come to expect that with AI but the impact it continues to have on how we work and complete tasks is so disruptive, even I find myself questioning my workflows more than a few times a week.

A few weeks ago, I sat down to record a walkthrough on creating skills using the skill editor in Claude.

All I could keep thinking about was “never did I imagine I’d be sitting here explaining how to deconstruct a skill to artificial intelligence”, only a decade ago most tech I was using could only handle static automations.

Before we go on, let me explain what “skills” are in the context of AI, specifically LLMs.

As AI becomes more ingrained into life and work, it’s only natural that it’s evolved to learn how to mimic our workflows to complete tasks.

That’s basically what skills are.

A file with a set of instructions, context, and guidelines that enables LLMs to extend their capabilities beyond the usual chat, that most of the population utilise daily. These are written and exported in a format called Markdown.

Don’t worry, it’s nothing complicated.

A Markdown file is a plain text document that uses a lightweight markup language called Markdown to add formatting. These files typically have the extension .md. Basically, they’re a package that LLMs can read much better than typical formats like docx and PDF.

You don’t need special software to create a .md file.

You can compile all of the contents AI needs to know in a standard word doc or Google doc. All you need to do is save/export as a .md file. Easy, I promise.

The content inside this Markdown file for an AI skill usually includes:

For Claude users, it will use that skill without being prompted to.

It knows to do this by reading the tasks you ask it to do and mapping that against available skills in your library. Pretty smart, right?

As a skill file is ultimately just a text file written in markdown. Any LLM that accepts file uploads or pasted text can use it as context. But, it won’t automatically activate and map it like Claude does.

If you take a snapshot of how L&D pros are using AI right now, the vast majority are typing prompts into ChatGPT, Claude or your poison of choice, and praying that something good comes out.

That's where everyone starts, but I feel like too many are stuck here.

Knowing how to craft clear and reusable skills that AI can utilise is going to separate you from the majority of people around you. There’s also an added unforeseen benefit here, and that’s forcing you to explain and think about how you complete tasks today, and what that workflow looks like end to end.

It might not have been something you’ve consciously focused on as you just do the task.

So, AI forcing you to do that feels like a good thing for your metacognitive capabilities. Plus, you have to continue to maintain that skill within AI’s system. It’s not a one and done event.

These and MCP’s (Model Context Protocols), that we covered in a previous conversation, are the combo that will enable you to do more with AI than just “chat”.

The TL;DR: Invest time in understanding how to craft skills for AI + provide context. You’ll build more leverage and efficiency with AI this way.

To start, identify a workflow that you do on repeat.

It can be anything, I’ve shared one of mine in the next section if you want to check that out to get your own creative juices flowing.

Now follow the steps below:

You may know I have a YouTube channel.

A lot of people do. I, however, know nothing about how to be successful on YouTube. I like making videos but building the packaging (thumbnails, titles) and distributing them (timing) is not my strong point.

This is where I ask AI to help me.

I write scripts, shoot and edit the video, and then I have a set of Claude skills (some might call these agents, but meh!) that produce video titles, write SEO/AEO friendly descriptions and advise on thumbnail selection. I built these skills through the best practices of YouTube strategists I’ve met and countless articles, and podcasts.

These enable me to extend my capabilities and Claude’s for something with meaningful benefit for my business.

You could do the exact same thing in your L&D work.

The beauty of skill files being built in a .md format is that you can take them anywhere because every LLM reads .md files. This is another good reason why it helps to build skill files. You can package them and take them with you to the next tool you use.

You don’t have to be held prisoner to any LLM if you build a library of skill based .md (markdown files).

Don't overthink it.

Pick one task you do every week that you can deconstruct into a workflow to use in your skill.md file.

Then:

Well done, you created your first skill.md file for AI.

Right now this works best with Claude (yes, I’m Claude-pilled), but you can upload markdown files to any LLM. I’d add them to your project folders, memory or custom instructions.

So, there we have it, folks. 

Your first entry, should we say, into the world of building skills for AI. This is by no means comprehensive, and if it is something that you want to get deeper into, let me know by hitting the reply button. 

I'm pretty big on this being a key skill for this year and probably 2027, but with all of these advancements, who knows what's going to happen?

- - -

P.S. Want to become an AI-Native L&D pro?…my next half-day workshop on helping you with this is coming up in June. I bring 10 L&D pros together to connect, learn and get a complete operating system for working with AI in your L&D role.

FYI: This isn't a tool demo or a prompting workshop.

→ If you’ve found this helpful, please consider sharing it wherever you hang out online, tag me in and share your thoughts.

VIDEO THOUGHTS 💾

In this quick tutorial, I’ll walk you through how to tailor your video overviews in NotebookLM, from choosing your format and language, to setting a visual style and telling the AI host exactly what to focus on.

Enjoy 😊.

PS… If you’re enjoying the newsletter, will you take 4 seconds to forward this edition to a friend? It goes a long way in helping me grow the newsletter (and cut through our industry BS with actionable insights).

And one more thing, I’d love your input on how to make the newsletter even more useful for you!

So please leave a comment with:

I read & reply to every single one of them!

Here’s four ways I can help: