Most people think building a Claude Skill means opening a code editor. Wrong.

A Skill is a set of instructions, written in plain English, that Claude keeps and reuses. If you can talk a contractor through a job, you can build one.

I built the entire AI layer of my business on ChatGPT. Then ChatGPT 5 arrived and I hated it. Properly hated it. The writing turned to mush, the responses felt lazy, and I made the call to move to Claude.

So when OpenAI rebuilt the whole thing this year, new models, a new desktop app, an actual agent called Work, I expected more of the same dressed up differently.

There's a box in your Claude settings that most people fill in once, badly, and never look at again.

It's the profile instructions field, and it's the only thing Claude reads before every single conversation you have with it. Not the ones in a Project. Not the ones in a Skill. All of them.

I hit my Cowork usage limit at 11am on a Tuesday 😬.

I hadn't done anything dramatic. No marathon automation build, no six-hour hyperfocus spiral (which, for the record, is absolutely something I'm capable of). It was just a normal content morning…a few emails, some system work, a couple of tasks. And then: limit reached. Come back later.

Most people I talk to are running Claude on one layer.

Open a chat, write a prompt, get something useful back, copy it into wherever it needed to go. That is genuinely useful and I am not going to be sniffy about it, because it is where I started too. But every session begins cold, and nothing happens unless you are the one who starts it.

The first time I set a job to run on a schedule, went to bed, and woke up to it already done, I sat there staring at the screen like someone had broken in overnight and tidied up (not even a joke) 🤯.

I was genuinely excited because it showed me that for the first time it looked like work was going to come off a solo founder's plate properly, not just come off faster.

Cowork is the bit of Claude everyone gets excited about. It's also the bit most people give up on because it turns up knowing nothing about you, and nobody really explains that part.

So here's what's actually going on. You point Cowork at a folder on your laptop and it can genuinely see everything in there... your client files, the half-finished sales page, three months of content, that spreadsheet you keep meaning to sort out. It can open all of it, change it, save new work back. That part is brilliant.

Here's a confession: the Skills I never got around to building were always the annoying ones. The ones where if I wanted to truly explain them you’d have to stop and record a whole Loom, because there’s no easy way to just write them down. Downloading transcripts, shuffling files between six tabs, that kind of thing. I knew they'd make good Skills. I just... never did it.

A few dozen times a month, someone types some version of "how to start a business with Claude" into Google and lands on a post I wrote for people who already have one. Right question, wrong post. Sorry about that 😅.

I’ve come to realise that there’s something almost every Claude guide assumes: that the business already exists. Existing offers, an existing brand voice, a folder of content to load into a Project. If none of that exists yet, that advice is standing one step ahead of where a lot of people actually are.

Yes. (And I know that's not what the article you read three months ago told you. 👀)

If you searched this expecting a no, you're in good company, because that was the honest answer right up until recently. Then it changed. ChatGPT has Skills now, and here's the part nobody expects: they run on the same open standard Anthropic built for Claude. So a Skill you make in Claude can move straight over to ChatGPT, and one you build in ChatGPT can come back the other way.

Everyone keeps asking me the same thing since ChatGPT Work landed: am I switching? The better question is how to use BOTH.

Here's what actually happened, and why it's bigger than either tool. 

Old ChatGPT answered your questions and drafted text in a chat. You still took that text and went and did the actual job yourself.

I was in Claude the other day trying to connect Metricool, and it wasn’t there. Not every tool you run your business on has a listing in Claude’s official directory, and for a while I assumed that meant copy-pasting data manually forever. It doesn’t. If a tool has an MCP server, official or third-party, you can add it yourself as a custom connector, no code required.

Try to remember the last time Claude told you one of your ideas was bad. Not gently improved. Not "here's a stronger version." Bad. A flat "I wouldn't do that." Most people, when they go looking, come up empty. It has definitely happened to me before, but I can count on one hand how many times, and I use Claude A LOT.

Your week as a solo founder is a stack of small jobs that all need doing and all land on you: sort the folder that's become chaos, write and create your content, turn a page of messy notes into a proper document, work through the replies sitting in your inbox.

Saved prompts still need you at every step. Here’s how solo founders turn their best prompts into AI systems that run whole functions; no code required.

A few weeks ago, I typed one sentence into a fresh chat: “I want to write a blog post about the difference between a Claude Skill and a Custom GPT.” Nothing else. 

Claude told me it was triggering two things: 

1. My blog writer Skill

2. My brand playbook Skill

A while back I signed up for an AI course that was well structured, well produced, and taught by someone who clearly knew what they were doing. I firmly believe that there’s nobody who knows all there is to know about AI and business, and I’m so happy to learn from those who are further ahead than me.

Most AI use case content is written for a different kind of business than yours. And I hope that by me saying that, you can finally exhale.

The content is usually for bigger businesses with bigger budgets, departments, tools most solo founders would never pay for, and the problems that come with all of that. So the use cases on offer are things like proposal automation, lead scoring, sales call analysis, and meeting summaries. Real problems, for that kind of business.

Back before connectors were a thing, I was using Claude to help me write client emails. I had the context in my head, typed it all out, Claude drafted the reply, and I copied it over to Gmail myself. Every time.

We tell Claude things all day. Our offers, how we like responses formatted, the project we’re in the middle of, and the way we want to be spoken to. And it’s easy to assume all of that is going into one ‘big brain’, so that wherever you open Claude, it just knows you.