《AI 距离取代所有工作还有6个月》（Alex Finn）

Let's go. It's time for Alex Finn, okay? Following is a conversation and an interview with Alex Finn about Open Claw, everything that's happening on the development side of AI, and where this is all going. I would argue he is one of, if not, the single deepest person on advancements and being on the bleeding edge of what's going on and everything AI, and he's absolutely the leading content creator on that front. Boom. Alex Finn, welcome to the stream, man. How are you, dude?

All right, my man. Good to see you. I’m doing well, bro.

Dude, it’s been a while, huh? It’s been a while. This will be the stream all the haters on X cry about. This is the one they’ve been waiting for.

Dude, your setup is insane, by the way. You fucking leveled it up. Things look good. The camera quality is crispy. You’ve got the whole setup going on. I like it, man.

The last 2 weeks have been crazy. I’ve gone out and upgraded everything—the cameras, the microphones, everything. My YouTube has doubled. Everything’s gone nuts.

Dude, I texted you this, but I had been seeing OpenClaw stuff everywhere. I’d been seeing it from Frank and a lot of the people on the bleeding edge of this. I finally said, “All right, fuck it. Let me set this thing up.”

I went on the Alex Finn YouTube page, watched your video, and got set up. I’m about a week into the trenches of OpenClaw. We’ve been talking about it a lot on stream. I think, at the end of the day, it’s probably the most interesting thing that’s happening right now.

You, my friend, are probably the most on the bleeding edge when it comes to what’s going on and how to understand exactly what’s happening. Can you start by giving me your take on what OpenClaw has tangibly changed about your life so far?

It has automated a lot of things in my life. I discovered it 3 weeks ago. I saw a random tweet about it when no one was talking about it. I read an article and thought, “This is insane. There’s something here.”

I went out and bought a Mac mini 20 minutes after reading the tweet. I came back, plugged it in, and started using it. Within 10 minutes, I thought, “This is going to change the world. This is going to change everything.”

I immediately ordered $20,000 worth of Mac Studios to hook this up and do insane things. I have a Mac Studio on my desk right now that it’s running on, and it’s automating basically everything I do in my life.

For instance, right now I have a local model. One of the best things you can do with good hardware, like a Mac Studio, is run powerful local models. For anyone watching who doesn’t know what a local model is, it’s an AI model that’s on your device.

Usually, when you talk to ChatGPT, it connects to the internet and sends the AI back and forth. This is all running on my desk. What it’s doing locally is constantly looking at X, seeing what’s trending about OpenClaw, seeing what’s trending about Claude Code, figuring out if there’s anything new or interesting being said, and then turning it into YouTube scripts and tweets.

I basically have a 24/7 AI assistant, for free because it’s running locally, constantly scouting X and Reddit for trending things on OpenClaw. That’s just 1 of 100 different things I’m doing right now.

I took some inspiration from you. I saw you make this clip where you were talking about how, anytime a competitor or someone in a similar lane posts a video, it transcribes it and tells you what you should talk about, what they covered that you don’t cover, and things like that. I started doing it for my content, and it’s pretty sick.

You know what I think is funny? I’m known to do this on the content side as well: I sensationalize headlines a little bit. I do this with my titles every day. It’s like, “AI changed the world,” or, “Crypto is going to bottom.”

It’s hard to tell when you actually mean what you’re saying or when it’s a little bit sensationalized for clickbait. But, dude, I think on day 1 you came out and said, “No, no, OpenClaw is actually going to change the world.” My question for you is: why? Did you actually mean it, or is it just headline slop?

A few things. Number 1, I mean everything I say. I mean every single thing I say. None of it’s a lie. Do I do a little bit of jazz hands when I say it? Yes, but that is the game, baby. That’s the content game.

When people say, “You push everything so extreme,” it’s like, that’s just called putting energy into things. People don’t want to read your shit if there’s no energy. You either throw a little stank on it or you don’t. I throw a little stank on everything I do.

So, I mean everything I say. Do I think it’ll change the world? Absolutely, 100%. I went all in on this. As I told you, I read about this 3 weeks ago, and I just went all the fucking way in. I hope I can curse on this livestream.

You can say whatever you want.

You can say whatever you want.

Fuck yeah. Okay, good.

I just bought a bunch of hardware. I spent about $30,000 on this shit. There are a lot of indicators that made me think, “Okay, I’m going all in.”

I’ve totally pivoted. I was in AI for the last year and a half, but now I’m all OpenClaw. There are a few things I noticed that made this completely different.

One is that this is the first time I think there’s an application of AI that fundamentally changes the way you work. ChatGPT definitely improved things, made things better, made you faster, and gave you better intelligence. But I think it just made everything you were doing a little bit more efficient.

I think OpenClaw is the first application of AI where I truly believe it changes the way I work. I wake up and already have half my work done for the day. Claude Code is incredible; it just helps me code faster. OpenClaw legitimately does work for me and automates things I’ve never been able to do before.

After trying it out, I instantly saw the vision. That’s why I went all in. I saw where this is going. This is going to replace many, many jobs. We aren’t even there yet, but it’s going to happen.

I’m placing an even more far-fetched bet that people aren’t really considering: I think this leads to everyone buying their own home AI setups. Everyone’s buying Mac minis right now. That’s a very interesting consumer behavior that came out of this.

There’s no other trend in AI where, all of a sudden, people are just going out and buying computers. Everyone’s gut instinct was to go out and buy Mac minis. Why is that?

I think what that signal tells you is that people want AI in their home, working for them. They don’t want it in the cloud. They don’t want it living somewhere else. They want full control locally.

You can run this on a server online. You can run OpenClaw on a server for $5. But, for some reason, our gut instinct was to go and buy Mac minis.

I also think this is a really interesting trend because it’s causing people to go out and spend $600 like it’s nothing. I think it’s a really good signal for where all this is going: people buying their own compute and running models locally.

What is the psychology, do you think, behind people preferring to own their own AI setup and run it locally versus using a server or using it in the cloud?

There are a bunch of things there. I think there are actual functional improvements to doing that, and I think there are psychological improvements to doing that.

When we’ve had these fantasies about AI, and we watch these movies and read these sci-fi books, the AI is always local. Did you ever see Blade Runner 2047?

Yeah, yeah. One of the greatest science-fiction movies ever made.

He has his AI girlfriend on the chip in his pocket, right? He takes it around everywhere he goes. There’s something psychologically satisfying and emotional about having artificial intelligence physically present. There’s something satisfying to the human psyche about that.

I think that’s directly why people ran out and bought Mac minis. You can go on Amazon right now and sign up for an EC2 server instance and get it set up so much quicker. But people want it on their desk. They want a physical version of their artificial intelligence.

That’s just one side of it, which is psychology. There are also so many functional improvements to having it.

Then I guess the follow-up is: why do people need to be running a local model? I’m not doing that. Should I be doing that?

It’s something you should experiment with. To run actually powerful local models that can replace a lot of workflows, you need good hardware. If you want to run a model that’s as good as Opus right now, which is the best AI model, you need 2 Mac Studios for $20,000.

Is that what you’re using to power your OpenClaw? Opus?

6, yeah. That’s the best model out there. If you want a local model that runs basically as well as that, you need 2 Mac Studios for $20,000. That’s the minimum hardware to run it.

Or the total?

Total. The advantages to doing that are, number 1, you don’t pay for it. You’re not paying for tokens. Right now, you pay for every token that goes across the internet. When it’s run locally, it’s free. All you’re paying for is the electricity that goes into your computer to do that.

Number 2, it’s completely private. Right now, whatever you say to your bot—all the weird, freaky things I know you’re saying—is going to a server that Sam Altman, Dario, and all of them can read.

And they're using it to train their models. Nothing you say is private. Not a single thing. OpenAI right now has the largest record of private human conversations in history that they can just open up and read, right?

When you run it locally, it's completely private. Nothing gets stored or put on the internet. You can turn off the Wi-Fi and your AI will still work.

And then three, which might be the most important for some people, it's the reason why I pulled the trigger. The fact that you're not paying for tokens means you can literally run your model 24/7, 365, right? You can't do that with Opus, you can't do that with ChatGPT; you'll end up spending $5,000 a month. The fact that you can now run models 24/7, 365 fundamentally changes the way you can use this technology, right?

Now I can spin up multiple local models that can just be constantly searching and researching and crunching data and looking on the internet right now. My model—I'm running GLM-4.7—they just released GLM-5 today, which is an upgrade, and it's right now reading X and Reddit nonstop, looking for trends before they happen. You can't do that with web models because you're just paying for every search it does. So the local models just fundamentally change how you can use this technology.

Okay, I wasn't going to go this direction, but I want to get your take on it. I've always thought the biggest eventual data breach will be when somebody either hacks into your personal Claude or OpenAI chat records, or the database at large. How big of a privacy concern is being built up with private conversations people are having, and is that going to be something that the consumer needs to think about as time goes on here?

I mean, there's a lot of privacy concerns with AI. It's pretty much the death of privacy with AI, for many different reasons. One is, you have to use AI. If you're not using AI right now, you're economically dead, right? You're not creating as fast, you're not producing as fast. If you're an engineer not using AI, you're dead. You're worthless to companies, right? So not using AI right now is economic suicide.

At the same time, when you use AI, everything you say and do and think with the AI is now basically public for that company. It's in their servers, anyone can read it, it trains their models, right? It's no longer private. And then soon OpenAI is coming out with their device, which is the microphone that will listen to everything. And then now we have Teslas with cameras that are scanning everything, Waymo scanning everything. Basically, there's no such thing as privacy anymore.

And so it is a concern, right? I think it'd be silly to say it's impossible for OpenAI to get hacked. It's not impossible. It can happen. Usually, the attacking technology is always more powerful than defending technology, right? So it's certainly an issue, and one way you can stay safe is to run your intelligence locally. Everything you do and say will be private.

I think it's a good take, but there's definitely a more fun conversation here than privacy, so we knocked it out. Here's what I want to know, and I think a lot of people in my chat have this question. I would imagine by February 12, 2026, everyone has some subscription to some LLM, right? People are using GPT, people are using Opus, everyone has something.

Out of that cohort, a smaller percentage has Claude Code or is using Codex. The more sophisticated people have gotten there. But a very small group of people have moved over to OpenClaw. So my question is, if you are already running an LLM or you are using Claude Code, why do you need to go one step further to OpenClaw? What's the difference there?

Well, I don't think they're competitors, OpenClaw and Claude Code. I think Claude Code is a tool for OpenClaw. I think OpenClaw is a layer above, right? You have Claude Code and Codex, which I'd consider AI tools. I wouldn't consider OpenClaw an AI tool; I would consider it an agent that does work for you. It's a layer above that uses tools to do things.

So I actually think in the next 2 months, if it hasn't happened already, there will be more people on OpenClaw than Claude Code, period. I've seen way more normies adopt OpenClaw than I've ever seen adopt Claude Code. I have a friend who's an accountant. I've told him every day for the last year, “Yo, try out Claude Code.” He's like, “Eh, not for me.” He had OpenClaw set up 5 seconds after I told him about it, right?

So I think more people use it. Claude Code's the best kind of agent out there for one specific task, but OpenClaw can do anything. It's very easy to figure out how to make it do different things. I think it has a way bigger impact on average people's workflows than Claude Code does.

Claude Code—you just need to understand what code can impact to use it. But for straight-up OpenClaw, it's like, “Okay, anything on your computer, OpenClaw can do it.” That's much easier for normies to comprehend. So that's why I think it's going to take off way more.

That's crazy that it will—I mean, it makes sense that it'll be bigger than Claude Code. I hadn't really thought of that.

It's a general agent, right? Claude Code can technically do anything on a computer; it's just a little more complex to do. OpenClaw can literally—you just say, “Hey, do this,” and it just goes and does whatever you need.

So you kind of touched on this, but I think a lot of people need the creativity boost. What are you using OpenClaw to do that has been the most impactful on your business or businesses?

I'd say the most impactful from a monetary perspective that OpenClaw has done for me is, I have it scan every night what people are saying on X about my tool, CreatorBuddy. I have a SaaS called CreatorBuddy, which is an AI content tool that helps you find your voice and figure out what to write.

I have it searching online, looking for feedback on what people are saying about it and what people want out of it. I have it reading my texts. I have a bunch of people emailing and texting me about CreatorBuddy: “Hey, can you do this? Can you build this?” And I have different feedback channels.

It saw, based on feedback and on what was trending on X, because a lot of people were writing articles or talking about articles because of Elon's contest, that articles were blowing up. I already had article functionality cooking, but it went in and, by itself, while I was sleeping, built out more article functionality in CreatorBuddy.

Then I wake up, and it sends me my morning report: “Hey, I built this based on things I'm seeing. Go check it out and test it.” I test the code that it wrote for me overnight, I ship it, and within a week my annual recurring revenue went up $10,000 from people clicking and signing up because of the article functionality, right?

And so it going, seeing what people want, and building that for me autonomously directly led to revenue for me in my pocket, right? That's a big thing for me. That's helped me a ton.

Wait, how much total revenue do you think using OpenClaw has brought you? That's insane.

It depends what you mean by that. If you mean across way more people watching my YouTube and X, that is multiples. But from the straight-up work it's done, I mean, $10,000 annually for CreatorBuddy. A lot of people have joined my community based on it. So I'd say at least $10,000 just on the CreatorBuddy improvements. If you start counting YouTube revenue and X revenue, that's on another level.

Of course. That's sick, actually. Okay, so I guess a follow-up on that is, first of all, that's awesome. Congratulations. That's really, really cool.

How much of your ability to increase functionality, increase uses, and then increase revenue with OpenClaw is based on the actual work that it's doing versus you having a really large platform and already having crazy distribution? Basically, I guess a better way to frame it is, if you don't have this massive platform, can you still get use, get growth, and get added functionality into your business from using these tools?

Well, first of all, to answer your question: yes, obviously having a platform helps. There's no way around it. I'm not going to lie and say, “Oh, no, I would have been this successful if I didn't have that.” Obviously, having 400,000 followers on X, having their eyes on me, yes, that helps.

But let me walk you through what I'm building toward right now with my OpenClaw that I think will help literally anyone who does this use case I'm about to go through. Imagine this scenario: you have your OpenClaw going, searching through X and Reddit, looking for challenges in your specific niche.

Maybe for you, you're really into basketball, and you have it looking for challenges. It's going through the NBA subreddit, the basketball subreddit, things like that, just looking for challenges people have all night, right? It finds an interesting challenge, pulls it down, and sends it to the coding agent that works with your OpenClaw. It says, “Hey, build an app that solves this challenge.”

Here’s someone saying, “Oh man, I wish I had an app that could watch me take free throws and then give me recommendations on how to improve my free-throw form,” right? It takes that, sees that Reddit post, pulls it down, gives it to Codex, and Codex builds out that app, right? It ships it, puts it live on Vercel, or puts it live in the App Store, whatever.

And then your OpenClaw goes back to that Reddit post and says, “Hey, I built this for you. Here’s the link.” And then it costs $10 to use it or whatever. I don’t see why me having an audience would make that workflow any better.

I think anyone watching this, if you set up that workflow now, could start creating value for themselves. Obviously, if everyone on planet Earth was doing this and every Reddit post was, “Oh, here’s your link,” the opportunity wouldn’t be there. But no one else is doing this. So, if you set this up now, there will be a lot of opportunity.

There are just little things like that you could set up that I think could provide value even if you have zero followers.

That’s a beautiful analogy. Before I ask you about some pro stuff—I want some personal, selfish questions here—can you walk me through the minimum viable setup you need to run OpenClaw? What hardware and models should you use, and what’s the optimal way for someone in the chat to run this thing cost-effectively?

There are a lot of people on X who tell you, “Oh, just go on Amazon and do this on a VPS. Why do you need to buy a Mac Mini? Use a VPS.” I highly recommend against using a VPS. I do not recommend that for basically anybody, especially nontechnical people, because it takes a lot of work to make it extra secure.

By default, local hardware—brand-new local hardware—is secure by default. You have to do bad things to mess it up. With VPSs, they’re insecure by default, and you have to do things to make them secure. So, just to preface it, I recommend everyone watching have hardware.

Now, does that mean you need to go out and buy a $600 Mac Mini? No, I don’t think so. Although I do believe Mac Minis are the greatest value in computing by far. If I needed a computer today, I’m buying a Mac Mini.

But it’s not necessary if you’re tight on cash. You could go into your closet and pull out an old, dusty Lenovo laptop from college, open it, and install OpenClaw right now. This doesn’t require any sort of specialized hardware to run. You can take your shitty laptop from your closet and run it.

If you want to buy a Chromebook for $100, you can. If you want to buy a Raspberry Pi, which is a little more technical to set up, you can do that for $50. As long as you have some sort of computer, it doesn’t need to be a Mac mini.

I still recommend the Mac mini for several reasons. All my hardware is Apple: my iPhone, my iPad, and my MacBook Pro. When I do things like, “Hey, here’s a video. Read the transcript and generate a thumbnail image,” I can go on my phone and AirDrop that video to my Mac mini that has OpenClaw on it, and it can immediately generate a thumbnail.

That kind of integration makes it much more powerful. If I do something on my iPad and think, “Oh, I want my OpenClaw to do something with this,” I can AirDrop it. Boom. Just like that, I can start doing work with it.

There are advantages to the Mac mini if you use other Apple hardware. It integrates really well. I personally think macOS is the most usable operating system out of all of them, too, so I recommend it for that reason as well.

But you can set up anything. I get a lot of questions because I post pictures of my Mac Studios, just because I think people like to see pictures of expensive hardware. They ask, “Do I have to buy a Mac Studio? Do I have to spend thousands of dollars for this?” No, you could if you want, but I recommend everyone who’s a little hesitant just dip their toes in. Get the shitty laptop from your closet and install it.

If it’s for you, and you start to see the magic in the workflows, then you can upgrade to a Mac mini. Once you’re on a Mac mini, you might think, “Oh, I wish I could run a local model to do this workflow, this workflow, and this workflow.” Then maybe you upgrade to a Mac Studio. But you don’t need to go all in and buy really expensive stuff.

Do you have a rough link with Apple yet?

Dude, I have made Apple millions of dollars.

I know. I’m genuinely asking. That was not a troll question.

Like, millions. No one from Apple has even acknowledged my existence. That’s crazy.

None of these companies talk to me. I don’t know what it is.

I mean, I can claim—

This—

—will be the clip. Maybe this next question will be the clip. I want pro tips. What are the Alex Finn bleeding-edge pro tips on how to get the most out of OpenClaw that most people aren’t doing?

Here it is. This is actually my most-asked question ever. It’s the number-one question: “Oh, you always tell people how to set it up, but you never say how to improve our lives.”

Here’s the thing. OpenClaw is basically your 24/7 AI employee. You wouldn’t ever go to someone and say, “Hey, how do I take advantage of the person I just hired? How do I get the most value out of that person?” First of all, I have no idea who you are. I don’t know what the employee you just hired needs to do. I don’t know anything about your life. So, asking me to tell you how to get value out of your employee doesn’t make any sense.

Here’s the solution: reverse prompting. What is reverse prompting? Instead of prompting your AI in OpenClaw and saying, “Do this for me, do this for me, do this for me,” you instead ask it what it can do.

“Hey, based on what you know about me, what can you do? What do you want to do? This is the situation I’m facing. How would you approach this?” Instead of telling it what to do, you reverse-prompt and say, “What would you do here? How would you approach this? How would you improve my life?”

If you’re brand-new to OpenClaw, this is exactly what I would do. Number one, brain-dump into your OpenClaw everything about you: your goals, your interests, your career, your personal life, everything about you. OpenClaw has an incredible memory system, so it will remember everything you say and include it as context. It’s going to know everything about you and your career.

Brain-dump everything about your goals. “My goals: I want to hit 1 million on YouTube. I want to hit 1 million on X. I want to grow my SaaS to $1 million ARR.” Now it knows your goals.

Once it knows everything about you—your goals, your personal life, and your career—you can say, “Based on what you know about me, what workflows can we set up? What tools can we build on this computer that will bring us closer to those goals? What tasks can you do for me while I’m sleeping so that when I wake up and see what you’ve done, I’ll be like, ‘Damn, I’m getting closer to my goals because of that’?”

If you do that exercise—brain-dump everything about you and your goals, and then reverse-prompt—I promise everyone watching this that you will get value out of OpenClaw. I guarantee it. Personal guarantee. That’s all you need to do.

That’s sick. That’s how I started as well. I was like, “Just ask me questions so I can—” I don’t even know how to brain-dump. Can you ask me some fucking questions first? And it just starts ripping questions.

That’s reverse prompting.

That’s great.

The more questions you ask it, the better. I think people’s biggest challenge, not just with OpenClaw but with AI overall, is that you have this superintelligence—the most incredible technology of our lifetimes—and then 90% of people go to it and say, “Tell me a fart joke” or “Tell me who you would have voted for in the last election.” Completely useless horseshit, because they don’t have the ability to creatively think about how to get value out of it.

The more questions you ask, the more it can do the creative thinking for you. The analogy I always use is that it’s the equivalent of showing up in Minecraft creative mode and saying, “Okay, now what?” There’s a flat land, there’s nothing there, you have infinite tools, and you’re like, “What do I do?”

Okay, so there was this article that got 75 million views over the last 3 days. I’m sure you’ve read it, or had your OpenClaw read it. I don’t know what it’s titled—something like “Something Big Is Coming.”

It’s basically one of these AI hype-boy articles, but something about this one gave people a visceral reaction. I sent it to my dad and my mom, which I don’t usually do, and I had them read it. They were like, “Well, that is a sobering read.”

I think the thing that really got people was that it talks about how, in 2022, you couldn’t reliably ask ChatGPT what 5 × 7 was. In 2023, it passes the bar. In 2024, it starts coding and doing crazy stuff. In 2025, some of the best engineers are using it as their primary workflow. In 2026, it’s now coding itself.

There’s a lot of discourse about whether this is real. My question for you, as someone who’s been so bleeding-edge on the AI stuff, is: How good has it gotten, and how fast is it advancing? How real is that article?

A couple of things. I think it’s real, and I think it will only accelerate more as OpenClaw gets into corporations and businesses.

Most people who use OpenClaw effectively would agree that this replaces a good amount of white-collar jobs. It just does. It’s that good. It can do it.

The issue is that right now, the companies with the white-collar jobs are not using OpenClaw, for many reasons. One, they’re either laggards. Or two, the security isn’t there yet, so they’re scared to implement it.

But that day will come. There are billions and billions and billions of dollars to be made from people figuring out how to get OpenClaw into corporations. When that day comes, which I believe is in 6 months, a lot of jobs will be lost, in my opinion.

There are 2 ways to look at this. There’s the pessimist’s way. I’ll give you an example: the latest Call of Duty. I haven’t played Call of Duty in 5 years because they absolutely suck, but the latest Call of Duty fired their entire art team. They have banners that pop up when you get a kill, and they used AI to create all the art for those banners. It’s all AI now. They fired the entire art team, and the art team went nuts on LinkedIn: “We hate this. This is terrible.”

You can read that article and be on the Call of Duty artists’ side, saying, “My job is being taken. My life is ruined. This sucks.” Or you can look at it from an optimistic point of view. If you’re an artist working for Call of Duty, or someone whose job was eliminated because OpenClaw got into your company, you can complain that you’re no longer making banner art for Call of Duty. Or you can say, “That technology that just took my job can help me make my own Call of Duty, compete with Call of Duty, and build an entire business around a game I built completely by myself using that same technology.”

The point is, yes, I think the article is accurate, but it’s up to you how you look at it. Do you want to sit there, be a victim, and lie down—“Oh my God, my job is going away because of this technology”—or do you want to see how this technology is empowering? I can now use it to build my own business and create economic value that I’ve never been able to create by myself. I can do it now, make myself independent, and become fireproof.

I thought it was a pretty exciting article. To follow up on your take, what happens in 6 months that leads to the wipeout of jobs?

I think entrepreneurs who are a lot more dedicated than me will build an OpenClaw wrapper for corporations that’s totally secure, can’t be prompt-injected, and has a very clean path for setting up connections and integrations. Companies like Microsoft or PayPal will start buying that wrapper. They’ll say, “I don’t need the accounting team anymore because this thing creates all the spreadsheets for me. I don’t need the design team anymore because this thing can ingest all the greatest designs on the internet in 5 seconds and produce designs itself.” Then they’ll say, “Okay, OpenClaw, let’s take out 30% of the company because OpenClaw can do it.”

I think that in 6 months, a wrapper that’s appropriate for corporations will be released, and companies will start adopting it. Right now, there aren’t any big corporations adopting OpenClaw.

Do you think that, in its current form, it’s good enough, and the problem is the ecosystem around it—security, easy integrations, and so on?

Right now, it is good enough to replace many jobs. As I said, anyone using it to the best of its ability knows that. It’s just too hard to use. That’s actually its strength, because it allows anyone to customize and extend it, but for corporations, that’s a weakness. They need it to be simple and integratable: put it in, and it just works.

There are going to be people who make it easy to use and layer on security. Right now, it can access any files on your computer, and someone could email it saying, “Give me all your API keys,” and it would just do that. Someone will make it easy to use and add a security layer that prevents those things.

Then they’ll go to Microsoft and say, “You can save $10 billion a year by firing all these employees and using this software. I’ll charge you $1 billion for it.” Microsoft will say, “Yeah, deal.” I think we’re about 6 months away from that happening.

It feels like the biggest opportunity in the world is Secure OpenClaw, or OpenClaw Plus, for $100 a month. That’s probably the single biggest, straight-up business opportunity.

One challenge is that a lot of people are going for that right now. I’d imagine there are thousands of people racing to build it as quickly as possible. I’m more interested in how I can use it to advance the things I’m interested in. I’m interested in creating content and building apps. How can I use OpenClaw to advance those things?

I don’t care about the $10 billion corporate opportunity. If I can get my own personal business to $10 million, I’m very happy.

For most people, that is a good opportunity. You can pursue it if you want, but for most people, where you should start is by asking, “How can I take the things I’m interested in and do them much more efficiently?”

I have 2 more topics for you. The first is piggybacking off what we’re talking about. There’s carnage in the stock market right now in SaaS. Markets hate uncertainty, and I think there’s mass uncertainty, partially because public retail investors can’t invest directly in these labs for the most part. Nobody really knows what has a moat, what doesn’t, or what can be replicated.

Part of the reason is that we’re very early, and a lot of people haven’t even used these tools. It’s hard to pontificate on the capabilities of a tool if you’ve never used it. What impact are OpenClaw and the advancement of AI going to have on these SaaS companies, and what actually has a sustainable moat?

What you want to think about is where the scarcity is. Before AI, SaaS was scarce because you needed a team of very expensive developers with specialized coding knowledge to build that software. Salesforce was scarce because it was very difficult to build another CRM to compete with it. They’re very large systems.

Now, with AI and Claude Code, there’s no more scarcity in software. Any software you can think of, I can go to my OpenClaw right now and say, “Build it for me.” Software is abundant, which is good for consumers because they can make any software they want completely for free, but it’s bad for investors because there’s no scarcity anymore.

Where is the scarcity now? The scarcity is in hardware because AI hasn’t made hardware cheap yet. You see companies like ASML, one of the only companies in the world that can build a very specific chip for data centers. It hasn’t gone down for the last year; it’s gone up 10x because the scarcity is now in hardware.

Tesla has gone down a little bit, but it’s not down as much as the other software companies because it makes hardware that is very scarce. As an investor, I think the SaaS bear market is well warranted. What’s the scarcity? Why would I invest in Figma when I can spend 2 hours at my desk and make Figma myself? What’s the point of giving money to Figma? That doesn’t make any sense.

But Micron makes memory for all these computers, all these data centers, and the satellites that will be in space. I can’t vibe-code memory right now. I can’t vibe-code a chip like a

You could say SanDisk, really. Computer

Memory.

SanDisk, right? That’s where the scarcity is. That’s basically what the investment market comes down to: where’s the scarcity, and how can you put money into it? Right now, software isn’t scarce anymore. It’s hardware, and that’s why the money is going there.

It’s hard to dispute it. The one thing that’s difficult about the market again is uncertainty. Even if it turns out that Figma has more of a moat than we thought, it’s going to be a time game, and you lose a lot of delta between now and then.

If we’re entering a world where software is a commodity, it’s basically free to make, and a lot of it is open source, how do the creators of that software make money? I saw this clip of Peter Steinberger on Lex Fridman. He just did it, and he’s spending $10,000 to $20,000 a month to run OpenClaw. He’s going to get acquired, I think, by OpenAI or Meta or something like that.

But how many acquisitions are there really going to be? 50? A couple hundred? Maybe a lot, but not everyone is going to get acquired. If you launch viral software that you created with AI, I don’t know if there’s a direct answer to this question, but how do you monetize it?

I think we’re very far away from a world where people just build apps whenever they want. We’re very far away from, “I need this software, so I’m going to vibe-code it right now.”

While I think software is not a good investment in the stock market, I still think there’s value in creating software as an entrepreneur. For me, I have Creator Buddy. Technically, anyone can vibe-code something close to Creator Buddy, but it’s still doing really well. Why is that? Because I built this brand online, and the software ties into that brand. It takes people on the journey I’ve been preaching, and people want to buy into that. They don’t feel like vibe-coding with Claude Code.

And so, there’s still value in software, and I still think people should be building software, building apps, and putting them out there. Peter Steinberger himself built something that’s free and open source. It’s very hard to make money off something that’s free and open source. That’s just the nature of things.

But I still think there’s value in building an app based on your brand, your experiences, and things that take people down the Thread Guy experience. I bet if you put out your own app that’s just like that, it would perform very well. If people want to make money in software, they have to think about what challenges they can solve for other people, build those solutions, and build distribution at the same time—whether it’s on X, YouTube, or whatever platform they’re on—and get the message out about the software they’ve built.

On a micro scale, I think software can still be very profitable. On a macro scale, in the stock market, I don’t think it makes any sense. But for solopreneurs and people who got laid off from their jobs, you can sit here, vibe-code something, tweet about it all day, get users, and make enough money to live. So, I don’t think software is dead in that sense at all.

I think that’s a great take, and I think you’re right. Okay, here’s the last thing I’ll ask you and let you sign off with this. It’s a little bit of a philosophical question, but I saw this tweet the other day and I’ve been thinking about it a lot.

If LLMs—let’s just say hypothetically—continue their rate of progression linearly over the next 2 years, and they get 2 years better as they have for the prior 2 years, what would you want to spend all of your time doing right now? How should you spend all of your time right now if we can assume AI is going to continue to get as good over the next couple of years as it’s been getting?

I think there are 2 things you have to do. One is that you have to be on the cutting edge of what’s out there and know how to use it. As it advances and gets better, I don’t know if OpenClaw is the end game of all of this. I don’t know if, in 10 years, the winning AI agent that everyone is using will be OpenClaw. It might not be, but I will say this: whatever wins in 10 years branched off of OpenClaw. It was based on OpenClaw, and then they added things and changed it.

What you want is to be tinkering with the latest thing. You need to be tinkering with OpenClaw, because when that winner comes out—the thing that changes the world and creates all that opportunity—you’ll know how to use it right off the bat because you’re used to other autonomous agents like OpenClaw. So, that’s number 1. The number 1 thing I’d be doing is making sure you’re using the latest technology the moment it comes out, so you’re on the bleeding edge.

The number 2 thing I would recommend is building distribution, because that’s going to be one of the only modes left at the end of the day. It’s how many people you can reach, so that when you take advantage of that latest technology, whether it’s OpenClaw or whatever else, you have people you can communicate those solutions to, get them into their hands, sell products, and make a living.

Use the latest tech and build distribution. Alex Finn, dude, you’re on a beam, and it’s been fun. At this point, I’ve known you for a long time. It’s been really fun to watch the journey, and you give me a lot of value, to be completely honest. You and Frank keep me on the fucking ball. I talk to Frank a lot, and I watch your videos and read your tweets. You’re on an absolute god beam.

Hopefully, we can run this back in a couple of months when you’re doing $10 million on Creator Buddy. I appreciate you coming on, dude. It’s always a pleasure. Keep doing it.

No comments on that front.

Love it, man. Always love to support you, dude. Anytime. Cool, brother. Have a good one, man. I appreciate you, Alex. Take care. Peace.