Ep. 32: WTF is OpenClaw?

ByMaestro
LISTEN TO THE EPISODE

The whole goal of this episode is simply to put OpenClaw, an open source autonomous AI agent that went viral in early January, on your radar, and make a case for why you absolutely should NOT install it. This episode covers what “open source autonomous AI agent” actually means, it’s two biggest draws, and the one massive downside that should give everyone pause: SECURITY.

Full Transparency

Real talk,  there’s a good chance you’ve never even heard of OpenClaw, which is exactly why I’m talking about it today. You absolutely do NOT need to know about this thing, I just want to put it on your radar so that if and when you do hear about it, you have a rough idea of what’s going on…and can ignore it.

Full transparency: I don’t use it, I don’t want to use it, I haven’t installed it, I do not recommend you install it, and I don’t even fully understand what it is or why there’s so much hype around it. But I’ve read enough to be aware of it, and I want to share what I know.

The AI Effect

A quick aside before we hop into the main topic for today.

The price of SSD (solid state drives) and RAM (random access memory) is through the roof right now. The AI effect is hitting consumers, well, consumers who want to buy memory.

My DJing sidequests had me searching for SSDs recently and I was shocked and appalled at the price.

How up are we talking?

  • RAM prices have seen massive hikes, 200–400% year over year in many segments.
  • SSD retail prices have been rising in many markets; some 2–3x increases.

Why? Manufacturers are opting to make chips for AI data centers instead of consumer products.

Worth noting, memory pricing has apparently always had a boom and bust cyclical nature (who knew?), and right now we’re clearly in the bad part of that cycle. My completely unqualified advice: If you’re looking to buy memory, don’t. Things likely won’t come down until 2027, but IMO it’s worth the wait.

WTF is OpenClaw?

Ok, on to the main topic. OpenClaw is an open source autonomous AI agent that went viral a few weeks ago in late January. Allow me to break down that string of nonsense words for you:

  • Open source: You can see and modify the code used to build it. For context, OpenAI, despite the name, and Claude, are NOT open source.
  • Autonomous: Works on its own.
  • AI agent: A computer program that acts on its own. (Yes, redundant to “autonomous,” but here we are.)

So what you need to know: OpenClaw is a computer program that acts on its own.

If you recall, we’ve been talking about agents and agentic AI for..a really long time nowI. Agents are AI programs that can do things. It is AI that takes action instead of just chatting with you. That right there is apparently one of the two big draws for OpenClaw.

The Two Big Draws

OpenClaw literally went viral because of its agentic capabilities.

Agentic tools like Claude Code and Claude Cowork can be used for coding, vibe coding, organizing files, organizing desktops. OpenClaw can do all of those things, (and I really gotta emphasize the vibe coding capabilities here because that’s what folks seem most excited about), but it can also do things like book flights, change your calendar, answer emails, send emails…but without you telling it what to do.

Literally none of this sounds appealing to me, but I digress.

The second big draw is how you communicate with OpenClaw: messaging apps you already use, like WhatsApp, Telegram, iMessage, Discord, and Slack. You set it up as a contact and then you text it like a person. The setup steps are much more complicated than that, but that’s the net effect.

Fun fact: OpenClaw is not its own LLM (large language model). It uses what’s already out there, such as Claude, ChatGPT, or Gemini. You simply connect it to your LLM of choice.

The Massive Downside

The massive downside? Security.

The only way OpenClaw can do all of those things on its own is to have access to all of your passwords, API keys, logins, and sensitive data. It leaves users vulnerable to attacks and hacking, which has already happened and continues to be an issue. 

OpenClaw basically gives folks the keys to your digital life. Immediately no.

Again, I do not recommend you install it.

The Hype Cycle

There was a ton of hype and excitement around OpenClaw when it rolled out, like, a TON, and it feels very characteristic of the AI world. 

Folks were so crazy over this thing that it was actually trending for a bit to buy a Mac Mini specifically to run OpenClaw, because keeping a Mac Mini on at all times (separate from your main computer) would allow OpenClaw to act like a true AI assistant and always be available. All you’d have to do is text it.

I don’t understand the hype, but I’m also not deep in the vibe coding, agent-building world. 

The big dogs, unsurprisingly, disagree with me, and on February 14th, OpenAI actually hired OpenClaw’s creator, Peter Steinberger. Specifics of the deal have not been disclosed, but it’s a very safe bet that they’re paying Mr. Steinberger a pretty penny to “drive the next generation of personal agents.”

Classic move, buy the thing that’s threatening to do what you haven’t done yet. It’s giving Zuck.

How I Used Claude This Week

Each episode I share a quick example of how I used an LLM that week.

This week I used Claude to learn about OpenClaw.

I’m currently deep in the weeds of building out and launching new messaging offers for my business, so there have been no AI sidequests. Not to mention, that Claude usage limit definitely makes me think twice before asking things. Honestly, if I’m not coding I haven’t gotten close to the limits, and I’ve yet to hit my weekly limit, but still.

Like I said in last week’s episode, I’m not mad about the fact that Anthropic has and enforces usage limits. I might come to regret that statement one day, but for now, it’s true.

Da Wrap-up

OpenClaw: You don’t need it, you probably shouldn’t install it, but now at least you know what it is.

As always, endlessly appreciative for you and your curiosity.

Catch you next Thursday. 

Maestro out.