A month ago, I was given access to Claude at work.
To help prepare for further roll out, client work, and grow my understanding of AI in general, I made a deliberate decision: I was going to use Claude properly. Not just dabble. Not just throw it the occasional prompt and see what came back. I was going to start at the beginning, complete the Anthropic training, and experience it like a user in a broader rollout would.

Why? Because I spend a lot of time talking about technology adoption, and I think there is something powerful about putting yourself in the shoes of the people you’re asking to change. So I did the training. I read the guidance. I followed the breadcrumbs. And then I started using it.

I genuinely did not expect what happened next.

Starting at the start (on purpose)

I have used Copilot for a few years now. Long enough that it’s woven into how I work, and long enough that, if I’m honest with myself, I had stopped exploring it. There were features I had heard of and never tried. Capabilities I assumed I “would get to eventually.” It had become comfortable, which is another way of saying I had stopped being curious.

So when I picked up Claude, I made myself a quiet rule: don’t shortcut the learning. Don’t skim. Don’t assume that because I know one AI tool, I know this one.

This is the bit that often trips up technology rollouts. We assume that confident users of one tool will transfer their skills neatly to the next. Sometimes they do. Sometimes they bring their old habits with them and never discover what the new tool actually offers.

I wanted to notice the difference.

From chat… to projects… to building skills (much faster than I expected)

Here is what surprised me. I thought I would spend month one in the chat window. Asking questions, testing prompts, getting a feel for the tone. A gentle on-ramp.

Instead, within a couple of weeks I was already:

• Setting up Projects to keep context in one place rather than re-explaining myself every conversation
• Experimenting with Skills to make Claude do specific things in repeatable ways
• Catching myself thinking “how would I package this so I could reuse it?” — which is a very different kind of thinking from “what should I ask?”

That shift, from consumer of answers to designer of how I work with the tool, happened far quicker than I anticipated. And I think it happened because I’d done the training. I had a mental map of what was possible, so when a use case came up, I knew there was a feature waiting for me rather than having to discover it accidentally.

Knowing what’s there changes what you reach for.

This is something I want to sit with as a change manager. So often we focus on getting people to “use the tool.” But there’s a meaningful difference between using a tool and understanding the shape of it. The training gave me the shape.

What this actually looked like in practice

Talking about Projects and Skills in the abstract is one thing. Here is what they have actually become for me.

Projects as a thinking partner — and a shared one. I built out a project area loaded with the documents I needed, and started using Claude to do rapid deep analysis and strategy work across them. The thing that genuinely shifted my work, though, was sharing that project with a colleague and progressing it together. Suddenly Claude wasn’t a private chat I was having on the side. It was sitting in the middle of our collaboration, holding context for both of us, and helping us build on each other’s thinking. That is a very different model of “AI at work” than I have had with any other tool. I hadn’t yet explored this with Copilot.

Skills as a way of building little experts. This one surprised me the most. I have built my own personal coach in Claude. I have built a change manager tool. Both of them help me kick-start thinking on work — sometimes to brainstorm when I am facing a blank page, sometimes to move through a challenge when I am stuck in my own head. They are not replacing my thinking. They are giving me a structured nudge into it. Honestly, building them was as valuable as using them. The act of articulating “what would a good coach ask me here?” or “what does a change manager actually do at this point in the process?” clarified my own thinking before Claude even responded.

I went in expecting a tool. I came out with collaborators I had built myself.

That sentence still feels a bit strange to write. But it is the most honest description of where month one landed.

The early adopters group: where learning really extended

Here is the bit I did not see coming.

I joined a group of internal early adopters, and the conversations in there have probably accelerated my learning more than any feature exploration on my own. People sharing what they’ve tried. People naming the things that didn’t work. People asking the slightly silly question that turns out to be the question everyone else also had. I talk about the research and try to drive deeper connection in my work, but it was much different to experience it while exploring a new way of working.

It has helped me:

• Get ideas I would never have generated on my own
• See challenges before I tripped over them myself
• Extend my thinking beyond “what can this do for me today” into “what could this look like across a team, an organisation, a profession?”

This is social learning in action, and it is a reminder of something I write about a lot — that adoption is rarely an individual sport. The tool is one thing. The community of people figuring it out alongside you is another, and arguably more important, thing.

If you are rolling something out without a way for people to talk to each other about it, you are leaving a lot of value on the table.

The unexpected gift: it made me better at Copilot too

Here is the loop I did not anticipate closing.

Spending a month being deliberately curious about Claude has made me deliberately curious about Copilot again. The features I had not bothered to explore (especially Agents – to me this sounded like something for IT nerds and coders). Or other capabilities I had assumed I’d “get to”, particularly Notebooks.
I am now poking at them with new eyes, partly because Claude has shown me what’s possible, and partly because the habit of exploring has come back.

This is something I want to flag for anyone leading change: the value of learning a new tool isn’t only in the new tool. It’s in what it teaches you about how you learn, what you ignore, and where you’ve gone on autopilot with the things you already use.

Sometimes the best way to refresh your relationship with old tech is to fall in love with new tech.

What I’m taking into the work

A month in, here is what I’m carrying forward — both for my own use and for how I support others through similar journeys:

• Training matters more than we admit. Not because people can’t figure things out on their own (they can), but because structured exposure surfaces capability they would otherwise never reach for.
• Move people from chat to creation faster than you think. The leap from “asking questions” to “building reusable ways of working” — Projects, Skills, shared workspaces — is smaller than it looks, and it is where the value compounds.
• Make space for people to build their own tools. Some of my best learning came from building a coach and a change manager skill for myself. There is something about designing that teaches you the tool in ways using never quite does.
• Build the community alongside the rollout. An early adopters group is not a nice-to-have. It is where the tacit knowledge gets made.
• Use new tools to refresh old ones. The skills don’t stay siloed. Curiosity is portable.

I have a feeling month two is going to be even more interesting than month one. I’ll keep sharing as I go.