Moltbook. MoltBot. OpenClaw.

A “Reddit for AI agents.” Bots talking to bots. A massive security breach. Headlines hinting at AGI and the singularity.

If you have been seeing this everywhere and wondering what actually happened and whether you need to care, this episode is for you.

In this conversation, Kyle James sits down with returning guest John Marcus III, someone he trusts to cut through hype and explain what is real, what is mostly nonsense, and what actually matters for go-to-market professionals.

We break down:

1. What Moltbook and OpenClaw really are, in plain language
2. Why this moment felt new even though the underlying tech is not
3. The real security and governance risks behind the headlines
4. Why AI orchestration matters more than chasing tools
5. What GTM professionals should do now, later, or safely ignore

This is not a demo and not hype. It is a guided walkthrough designed to help you shut the curiosity door, get back to work, and be better prepared for what is coming next.

Links & Resources

1. OpenClaw (GitHub): https://github.com/openclaw
2. Fractional Ventures: https://fractional.ventures
3. Connect with John Marcus III on LinkedIn: https://www.linkedin.com/in/jwmarcus/

Multbot, Multbook, Claude Claw, whatever this stuff is like blowing up everywhere.

And I'm like, I need a friend of mine who understands what the heck this thing is.

So that's where this whole thing took off.

People were just giving it its own environment, letting it have access to their machines and running wild.

Give it its own little computer.

You don't have to worry about it having access to anything else.

It's contained on that box.

So in San Francisco, none of the Apple stores had Mac minis for like a couple days because everyone just went out and bought them.

And then you go get your Claude Max subscription at 200 bucks a month and just.

Here's my social security and here's, my will and testament.

And you can imagine where that went for all of the AI influencers.

What I'm hearing you say is like, one, the big revelation of this is the orchestration layer.

But also...

To be honest, I had the same reaction a lot of you probably did.

What the hell is Open Claw?

What is Molt Pot?

What is this lobster thing?

And do I actually need to care about any of it?

This conversation is not going to be a demo, it's not a guided walkthrough, but it is going to be a conversation for go-to-market professionals to actually be able to take away, what do I do with this?

How does it help me?

And shut the door on the curiosity with real answers and how they can apply that.

But before we start, if you like this sort of content, if you want to see more of it, can you do me a favor?

Can you give that subscribe button?

Just give it a good pop.

give this episode a like.

And if there's anything that you want to go deeper on or questions that you have, leave a comment.

Would love to hear from you.

And I'm figuring all this stuff out too.

I'm just taking you along for the journey.

All right, John Marcus III, JM3, super glad to have you back on the podcast again.

It's been a while.

But this whole molt bot, molt book, clog claw, whatever this stuff is like blowing up everywhere.

And I'm like, I need a friend of mine who understands what the heck this thing is and can kind of break it down and A, explain it to me.

But like, if I'm going to learn it, might as well share it with the general audience here on the podcast.

So welcome back.

What is this thing?

Tell it, give us the DLL on it.

Yes.

Thanks, Kyle, for having me back.

And

And you reached out and said, hey, do you know anything about it?

I was like, I have strong opinions on this thing.

Love strong opinions.

So what is it?

Okay, so it started out being called Claude Bot, C-L-A-W-D, but then inevitably Anthropic sent a notice and they're like, hey, don't do that.

And then he had to quickly come up with a name.

The creator of this project looked to the community and they said, Molt Bot.

Like it's like it's molting, it's evolving, like as lobsters do.

That's where the lobster connect.

You know, I didn't even know that.

Okay.

Yeah, the lobster connection was earlier than that.

The emoji for the lobster was there first, and it was like Claude Bot, and then, you know, the lobster was the avatar, the mascot, if you will.

So, because lobster is molt, it became Molt Bot.

And then everybody said, that's a terrible name.

So I'm not making this up, but go look at the story.

But then they settled on Open Claw, C-L-A-W.

Because lobsters have claws, right?

Because lobsters have claws.

Yeah, but the idea of evolution is still in there.

It's kind of memes on top of memes on top of memes at this point.

Yeah, and they were all like a day long, so they didn't even have any time to mature into like proper memes.

But Open Claw,

which it is now or for now, but we'll see how long that lasts, is an open source project created by one individual, a successful entrepreneur who wanted, as a huge fan of Claude Code, the tool, he wanted to go and create something more multifunctional out of that.

So what he did was created this thing called Claude Bot, which is now Open Claude, as an agentic

tool calling, memory persistent, multi-channel, agentic assistant, right?

There's a bunch of words in there, we can unpack some of them, but...

So it's like a master agent to rule the sub-agents or mini-agents.

Yeah, even less so that, but more so using the agentic framework, which at the core, agentic AI calls tools, right?

It can use things.

maybe with this little lobster clause.

There you go.

But it can use tools to do things.

And that could be searching the web, it can be interfacing with a database, it can be writing its own code and then executing inside of a sub-environment.

And that's kind of, at least in my book, what differentiates agents from your general LLMs and kind of broader machine learning and AI.

So that's where the whole thing came from.

And it really was the first normie available

type of system where people could go and set it up and then interact with it over a text message or over Telegram or over Slack or whatever.

And most of the stuff was kind of built in.

You could kind of connect it to this, connect it to that, give it all your credentials, let it go nuts.

So that's where this whole thing took off.

People were just giving it its own environment, letting it have access to their machines and running wild.

Now, some of the memes that quickly came up there were people buying up all kinds of Mac minis.

which was interesting.

So in San Francisco, none of the Apple stores had Mac minis for like a couple days because everyone just went out and bought them.

And then you go get your Claude Max subscription at 200 bucks a month and just let this thing rip.

Give it its own little computer.

You don't have to worry about it having access to anything else.

It's contained on that box.

Go.

Oh, but go ahead and give it your e-mail credits and go ahead and give it access to your calendar and, you know, let it just go nuts.

Oh, and your credit card.

And here's my social security and here's, you know, my will and test.

And you can imagine where that went for all of the AI influencers.

This was the bait for them.

So every AI influencer that you can possibly find on YouTube, I mean, YouTube is like where a lot of them live, but you just go and look at these videos like, oh, I gave it like $10,000 and I gave it access to my computer and I let it go buy whatever it wants.

And

other people setting up multiple bots to kind of talk with each other.

So you have like an open claw and an open claw talking back and forth to each other.

And oh, they created a plan.

So that was how it got started.

And then it just got sillier from there because at the same time, the author of Claude Bot also created this Reddit like social media experiment.

That's the old book.

That's Multbook.

That's the one that everybody's like, oh my God, this is freaking out.

Like, what are these things?

Here's the robots talking to each other.

And that was the idea.

You'd be able to send the robots, you know, you send your agent to 1 URL and it would get the instructions for essentially how to operate.

And there's a key in there that, you know, it goes and reads the web, it gets instructions and it acts on them.

Super dangerous, tons of security implications there.

But the idea was to, hey, you have these things reach out and essentially be able to converse and interact and engage just like you would on Reddit.

Now we've seen that a lot of that stuff was actually put out by humans.

There's very little actual agent interaction.

So it was mostly spoofed and like I could never get to the bottom of that.

Some people are like, yeah, this is people writing this trying to be cute.

And other people are like, no, they're telling their bots and their agents to do specific things and it's kind of acting upon that.

Yeah, or you could just go tell your bot to go say the thing.

The signature's there, right?

You're coming in as an agent, or you're coming in as a human, and then there's a human aspect to kind of like observe it.

So anyways, it's just a big old silly social experiment.

But again, just ripe for the picking for people to, you know, slap headlines.

Oh, we've achieved AGI.

Like, it can do all of these things.

It made phone calls for me.

And for every one of those successes, there's probably like, you know, 20, 30,000 failures.

But that's where it came from.

That's where it's at.

It is a very active, active project.

It's asymptotic in terms of the amount of stars and the amount of users on this thing.

But yeah, that's where it is.

That's OpenClaw.

I'll make sure I put a link to that.

If you haven't seen that, that'll be in the show notes because it's just worth knowing that it's mostly fake and kind of BS, but it's wild.

But I think, but I think that was the, that was the like buzzworthy thing because

As you and I were kind of talking before, it felt like all this stuff was happening.

None of this stuff was necessarily brand new.

It was kind of maybe orchestrated together a little bit, but this was the first time that the public, the masses, or maybe even the people that are like AI aware, they heard a bird chirping, but they didn't really see the bird.

And then like they took the blanket off of the birdcage and like, oh my God, this stuff's doing crazy stuff now.

Yeah, and there's some realness to it.

So I installed it, ran in a couple different instances and environments and hooked it up to different things.

There is within the prompting, which is where most of the work, and even today, prompting is key to all of this stuff for now.

The prompting is very well done.

And the community's put in a ton of effort in being able to connect it to all these various tools.

kind of like an MCP, but this thing kind of can go on its own.

Yeah.

And be able to do all kinds of what feel very novel, like independent research and like digging through websites and clicking through forms and like adding stuff to a shopping cart.

We've seen bits and pieces of this, but it's never been in one spot.

So you think back to Operator by OpenAI and their whole like, you know, integrated browser experience and it was slow and it didn't do a very good job and it had to like take screenshots and everything.

now you've got something that's got that kind of tool for web browsing, but it's not trying to control your computer.

It's got its own browser.

And if you think about it, that cuts out the visual layer, it cuts out all the other stuff that humans need, but computers don't.

It just needs to hit the website, be able to get the information in text format back into the model.

So it runs a lot faster because it's optimized for that.

Yeah.

It just skips all the human steps.

And that, I think, is what was

very interesting to the general public is that it could do things that you thought, or if you weren't really deep into this stuff, you thought were only human tasks.

Like only humans can go and click through and do these things, and it's so complex.

And yeah, normally it would be very hard to set up a script or Playwright or Selenium or something like that to go and do all those clicks and navigate through these sites.

While if you have something that can do tool calling and it has memory and it can go back and review its own notes, that's a marked difference than what most people have seen before, which was, I need recipes for a rhubarb pie.

Okay, so what I'm hearing you say is like, one, the big revelation of this is the orchestration layer, right?

But also, people probably shouldn't go play with this stuff because like security, governance,

don't give it your credit card.

You probably don't want to give it access to your computer.

Like the Mac minis is a reason even the crazy wild kids did it because they wanted to put this thing in a little controlled environment, right?

So there's probably not anything here that we should be telling GTM professionals that, hey, you should go out and play and try this stuff.

Like just watch maybe?

Right.

This is a great opportunity to watch.

If you follow this stuff as heavy as we do, get a couple chuckles out of it for some of the

the over-the-top reactions that some of these people are having.

But it's really important to understand that what's going on automatically here is for GTM professionals, people who are working and go to market, it's a very specific kind of shot fired, if you will.

You know, what started off as kind of an interest in a project and is open source, right?

There's like no company backing this thing.

One guy built the whole thing.

again, with a huge community that's caught on to this and said, hey, this is really interesting, I can find this useful, that this combined system can do things that look a lot and could even be part of your normal day-to-day job.

It doesn't mean you need to go install OpenClaw and run it on a Mac mini and give you a credit card for it to go manage your calendar.

But if you don't manage your calendar,

or you're bad at it and your job relies on being prompt and managing your calendar well, the robots have beat you, right?

So now the machines are better at that part of your job.

So I see it as a wake-up call to the masses to say, you need to look at what you do for work at home and give it a real clear review

in terms of what the machines can already do.

And this is a good demo of the capabilities, even if it's not 100%, it just takes time and more prompts to be able to get it to 100%.

And if things can go and book dinners for you, or it can set up travel plans and go get tickets and compare prices, and you're a travel agent, that's a really stark reality.

to see something that can run on a couple $100 machine or in a cloud for a couple bucks a month.

DigitalOcean has a one-click installer now.

I tried that one.

I was like, oh, it just works.

Give it all your credentials.

Your knowledge work and kind of the stuff that you're doing is going to come under fire.

And this is a good opportunity to kind of look and observe and see what it can do and reflect on that and say, how much of my job is still uniquely me, uniquely human?

And I think this is the first time that

There's a really large swath of not just a professional community, but people's personal lives.

that can be, let's just say, like imprinted upon.

Sure.

This machine, like I can see this machine doing it for me.

And then you go, wait a minute, so why do I need to do it?

Wait, what do I do?

And it's just like existential crisis ensues.

Yeah, I get it.

The way I think about it, the way I've always talked to people, it's like, you're working with these things and you're no longer the journalist, if we're using a newspaper analogy, you're the editor.

And that's what I hear you saying too, like if you are that travel agent,

You still have to talk to the people to figure out where they want to go, but maybe some of the like, all right, I know they want to go to this place.

Agent, go out and find the best prices for this in this timeframe.

Give me back recommendations.

Give me reviews.

What is the travel agenda like?

You don't necessarily have to manually go do all those steps.

And probably the advantage of you can tell it to go do five of those and bring them all back to you, and then you can kind of review them, right?

You still need that expertise human layer on top of it, but the manual labor probably

gets to go away a lot.

Yeah.

A lot of the manual labor of knowledge work, for sure, that's cooked.

So everything from web design, certain kinds of graphic art programming, for sure, that's getting completely upended.

And if you were really good at spitting out lines of code and perfect syntax, that doesn't matter.

Machines beat you.

So instead you have to elevate, not from a developer, you have to think like a product, you got to think about the end user, you got to think about the constraints around that, what the frameworks can and can't do, because that's not necessarily where a lot of these, even the more advanced kind of coding assistants like cloud code and cursor code and all that, or cursor CLI, they're not great at that yet.

Now there are things, there's the GSDCC framework that I've been messing with,

And that's a meta-prompting framework that will go through and kind of handhold you through breaking down an entire product.

And it's by Tashez.

He's an electronic artist who got into making apps.

And he's just like, here I am.

He did not write a line of code.

But again, he's using something, I think cloud code in this case, and a whole set of prompts that he and his community developed to ask the questions of you, the person who wants to make things,

and all the details and it goes and it does the research and it does the comparison.

So even the kind of management and kind of human expert on the topic, that's a threat, being threatened too.

Sure.

Because now you have systems like this GSD framework that will go and just interrogate you over and over in a loop until it's satisfied that it has the things that it needs to go from a concept to

a research document.

It'll then go and crawl the web.

It rips through tokens for sure.

And then it comes back and it says, I need clarifying statements.

It'll ask you that right into prompts.

And then it'll go through design, and then it'll go through implementation, and then it'll go through testing, and then it'll go through validation.

And it does it all on its own.

What I'm hearing you say is like everybody needs to like think a little bit more like an entrepreneur and less like a

figure out what the idea in creative and how to execute on it.

Well, you don't even worry about the execution as much.

It's like, can you be the creative spark now?

Yeah, the entrepreneurship is an interesting angle.

And I think that word, as overloaded as it is, is probably an important one to go back and study.

So if we think about what we do in our day job, you can have a very specific role and very specific requirements as part of your JD.

but you have to think, what am I converting from resources to finished product?

If I'm a knowledge worker, I'm converting ideas and client requirements to creative copy, to design works.

If I'm a developer, I'm taking that PRD and I'm turning that into bits and bytes that do what they're supposed to.

So

going back and kind of taking a hard look and saying like, oh, I'm a people person.

You're going to have to be more specific than that.

If you're in sales, you're taking raw knowledge of prospects that are out there in the market and turning those into human conversations that extract information that you need to see if they're fit for the thing that you sell.

So I think it's a good opportunity, even as, I don't want to call it a toy, because I think OpenClaw has got the potential to do a lot of stuff, especially if

those things are soaking up a ton of your time.

But if it's going to give you back that time, you got to make sure that you're doing the most that you can with the time you get back.

What I'm not hearing you say is this is definitely not AGI.

No, AGI is, the billionaires love the AGI thing.

They love saying AGI.

But it's also, I think it's also important to say like,

I don't think I'm hearing you say either, like, this is not helping us get closer to AGI either.

No, and I think that where we're at, there's some version, again, depending on whose description you use, there's some version of AGI already here.

And it's not going to be like, oh,

we have AGI and all of a sudden, bing, it's here, right?

It's Siri 5.0 or something like that.

This is a smearing of like technology and processes that are eventually going to get us to the point where some combination of tools can do what would give us some general intelligence style effects.

Now, what does it do, right?

What do you plug into it?

Well, as soon as we...

Remember the one thing I said, don't plug Skynet.

You're like, don't plug it in, right?

You'll get Skynet.

And here we are.

We're plugging it into all kinds of stuff.

So this is what I said very early on, and I know we've had some text messages about this.

Whatever you do, do not plug this thing into the real world, and people are doing it.

Do not give it the launch codes.

Don't do that.

Don't give it the launch codes.

Don't give it your credit card.

Don't give it access to your local machine.

Why?

Because

Is it going to go and become its own manifest and decide that it's going to go and take over the world?

No, probably not.

I hope probably not.

Probably not.

We don't know.

Probably not.

I mean, you know, you go back and you look at Oppenheimer and like his calculations were that it probably won't incinerate the entire Earth's atmosphere, probably.

But I can't guarantee you that.

So I got to always hedge on these podcasts. So

I would say the utility of a general assistant is much closer and I think we're seeing the first glimpse of it. And I've built a couple tools in similar fashion with memory, access to MCPs for clients of my own and then for myself to just stay on top of things. And they're generally useful. So again, it kind of depends like

Does it need to iterate and be able to improve itself? We've already got systems that can do that. Does it need to be able to interact with the outside world and do useful things that have economic value? We've got stuff that does that already. It doesn't need to be at parity or better than humans. We already have that. So I think the AGI discussion is muddy at best. And if there's anything to be

seeing from the people who are using AGI in their kind of day-to-day language, it's all CEOs of kajillion dollar companies, right? Yeah. It's marketing.

And it's funny too, because now you've got me remembering, like it's been a few years since we talked about it, but you remember the Turing test was everywhere for decades? Like, can we get something, a chatbot or whatever that could convince this, can it pass this test that people think it's real? Well, we did that. We did that some number of years ago. And this is like another one of those steps on that process. Like, yeah, it happened a few years

instead of decades later, but it's not, it's not to the point where it's like being the creative spark, as we talked about.

Yeah, nobody cares about the benchmarks. Nobody cares about SuiBench being, you know, one-tenth of a percent higher than the other ones. People are settling into whatever models suit their needs or have the, you know, vibe or attitude or the workflow that suits them. And then kind of these giant companies are going and essentially making opinions on where they want to be in the market. So you look at the Super Bowl ads and clouds, you know, very much like

we're not doing ads, right? They've doubled down on the code space heavily. And that's, you know, cloud code and the latest stuff with 4.6 Opus, its capabilities, how it's trained. And you go into ChatGPT and you go in your settings and now there's a little toggle for your ad preferences.

Yeah.

So that hit quick. So now you're saying, okay, well, I like this AI over this AI. I think back to when,

before dirt was invented, when you had a favorite search engine, and certain search engines were better than others at certain things. And then you had the meta search engines like Dogpile, and then you had the stuff that was a little bit more free text, like Ask Jeeves and stuff like that. In the end, it didn't really matter because search is search. And it all kind of became, you know, Google gobbled everything up.

But there's kind of like an expectation as to like what good search is that I can kind of get it on the first page. And it doesn't have to be the first, second or third, but like my result's going to be there somewhere. If I put in an address, it should open up a maps application and it's become second nature. So there's the skit by, I think it's Louis CK. He did the whole skit on flying.

Oh, I love that one. Yes. Flying across country, everybody's busy. Yeah.

You are flying like a bird through the air. And I think that's where we're at. So we've become a little numb to the novelty of the tech that we already have, and then you see the next step, and then people are like, oh, I can never go back to looking at my own calendar. But that might be the thing. The interface of two, three years from now might just be that text box.

Sounds like the big takeaway for everybody though is like, hey, this was cool, enjoy it. You need to start thinking about how you're going to start orchestrating all this stuff together.

the different subtasks, sub things that you do in your job, how could you start thinking about setting up ways, and I guess it's really building your context and your documentation so that these things can start taking over for you, so you can start leveling up and doing more creative things. It's coming. It might not be here yet, but be ready.

This is the wake-up call. I know that everyone gets alarmist about it, but I think that this is a really good example to

start to have the conversation with yourself about what things that you do in your job are unique to you. And if it's something that is repetitive and is easily done by the LLMs, you should give that up. And then you should try to think about ways that you can leverage, maybe not Claude Bot, but maybe it's a prompt. Maybe it's a text file that you set up and you say like, Hey, every single time I've got to write

of my TPS report, take this chunk of information, put it in a TPS report. So stop doing the TPS report, filling out stuff, and then go back and say, actually, my job's in negotiation or customer success or marketing or messaging, and figure out which parts of your job are going to go away because it's not, there's a lot of people say, the robots aren't coming for our jobs. It's other people who are good with AI that are going to take it because they don't have to spend any time

managing their calendar or their inbox, or they don't have to type out. It doesn't matter how fast you type. I can speak at 220 words per minute. I can't type at 220 words per minute.

Wow, look at that. It's a permission to slow down with all this. Awesome. Well, give everybody your lat long. Tell them how to connect with you. Tell them a little bit about Fractional Ventures and how they can help you.

Absolutely.

Fractional Ventures is the firm that I run. We help businesses with rev ops, AI, marketing, customer success, sales. We're really helping companies that are turning corners, that are going through transformation and need a hand. So we have fractional executives that we can plug in and we do all of our own work. So I'm doing the AI, I'm building the AI, I'm doing the rev ops, I'm sitting there and managing the Salesforce and HubSpot instances. We are accepting applications for our March and April classes.

If you are going through one of those go-to-market challenges, reach out to me, just fractional.ventures. And it's really simple, just john at fractional.ventures. Everybody knows my email.

And if you haven't picked up on it yet, he loves challenges. So if you've got somebody that like, I got a gnarly thing we want to solve and I don't know who can do it, JM3 might be your guy.

This looks dangerous and our business relies on it. Give me a call.

Give me a call. It could be an interesting conversation.

Talk is cheap.

So that's what's going on with me. And aside from that, I'm based in Boston, so we are freezing our rear ends off. We might, if we're lucky, get to the freezing mark, but not too hopeful yet. So if you're watching this in the middle of summer, you'll be with the other New Englanders complaining about how it's too hot.

Here we go. Awesome. John, thanks so much for joining me again and everybody out there.

Start orchestrating, keep learning, and until next time, keep growing. Take care.