If you’re of a certain age, you probably remember Leeroy Jenkins, the World of Warcraft player who became popular after a viral video in 2005. From Wikipedia:

In the video, having been absent during a discussion of a meticulous battle plan, Leeroy returns and ruins it by charging into combat while shouting his name as a battle cry.

Even if you haven’t seen the video, you’ve likely seen it referenced in pop culture, news, or politics. It’s come to mean everything from not taking things too seriously, to the perils of not planning (or even thinking) enough.

I find myself thinking about it a lot lately, and not just because it’s almost drinking age here in the USA. It’s because every time a new AI tool comes out, it seems like the internet goes all LEEEROOYYYY JEEENNKIINNS on it, claiming it’s the greatest thing to ever be invented. That it will change our lives as we know it. That if you’re not using it, you’re doing so at your own expense.

But my friends: all that shimmers is not gold. Sometimes it’s just a shiny.

Is OpenClaw the Next AI Savior?

Let’s take a look at the latest cause of everyone’s shiny object syndrome: OpenClaw.

Before we go on, I want to make it clear that this is nothing against the developer, Peter Steinberger, or Federico Viticci (whose article I’ll reference in a second). Steinberger has built something legitimately cool, and Federico’s whole job is experimenting and building stuff, which is ultimately where I’d like to be.

This is about the hype around pretty much all AI tools. OpenClaw (née Clawdbot, Moltbot) just happens to be the latest.

I barely have any idea of what it does, to be honest with you. I read about it on the website and in this comprehensive MacStories article. It seems like it can do whatever you want it do to…though I find those claims to be dubious. A lot of AI tools are guilty of making that claim and then not delivering…including Apple with their promised features of Apple Intelligence.

One reason I suspect the site barely says anything about what it actually does is this: it can connect to virtually any tool you’d like, but it takes a herculean technical effort to get it up and running.

You need to have a local server (or be dumb enough to run it on a publicly accessible server, which has devastating security implications). Then you need to be familiar with the command line. You need to know how to actually connect the tools. Oh and choose the LLM and the chat app.

You basically need to build the house to live it in.

There are Cool Experiments Happening

Now…once you do that, it seems you can do some cool stuff…at least in the short term. From MacStories’ Editor-in-Chief, Federico Viticci:

Every morning, [OpenClaw] sends me a daily report based on my calendar, Notion, Todoist, and more. It also sends me an audio version of the report that features silly artwork created by Nano Banana each day.

I say in the short term for few reasons:

• The MacStories article is 2 weeks old, so this is definitely more an experiment than a de facto way to do things.
• The software itself is 3 months old and has changed its name 3 times in as many weeks.
• Viticci has already made changes to his setup.

Since the article was published, Viticci mentioned on an episode of Connected:

because my article got so popular… I got very concerned about prompt injection attacks. I made the call of disconnecting [OpenClaw] from two integrations, specifically the Google Calendar and the Gmail one.

I strongly recommend the entire discussion because it’s a good, insightful one. Viticci has definitely embraced AI more than I have, and I honestly think he’s doing it in a really good way. He’s pushing the boundaries of what’s possible while not replacing himself in his creative work.

You Don’t Need to Try Every AI Tool

Again, my point isn’t to dump on any person or tool mentioned here. It’s simply a word of caution: don’t go charging into every AI tool, Leeroy Jenkins-style, like it’s going to change the world.

A great way to protect our time as solopreneurs is to not try every tool that comes across our screen. When I was in software development, and a coworker or contractor would recommend a brand new framework, I’d ask a simple question: “Are we doing this because it looks cool, or because it can do something our current tech stack can’t?”

I would always want a specific use case if we were going to upend our entire development system. And if they couldn’t provide one, we’d stick with what we have — the stack we’re all familiar with.

When it comes to new AI tools (or really, any new tools), your approach should be the same: look at what it does, create a specific use case, and see how well it does the job.

It could be that OpenClaw really delivers on the promise of AI agents. But it could also be that you spend a bunch of time building a house, only to realize the foundation is on quicksand.