#571 Scaling With Intelligence: Building an Autonomous Business With Amos Bar-Joseph

Mehmet: [00:00:00] Hello and welcome back to an opposite of the CT O Show with Mehmet today. I'm very pleased. Joining me Amos Bar Joseph, his founder and CEO of Swan Amos. I was telling you before we start the recording, the way I love to do it, I don't like to steal much from the show. And, you know, uh, steal from, you know, all the time that my guests want to have on the show also as well.

So this is why I'm gonna pass it to you. Tell us more about you, your background, your journey, and what you're currently up to. And then we're gonna dive into what you're currently doing. Today we're gonna talk, of course, about building autonomous business, humans operating in their zone of genius. And, you know, a lot of other topics, of course related to GoTo market and to AI also as well.

So the floor is yours. 

Amos: Amazing. Happy to be here, Mehmet. Thank you for having me. Um, yes, as you mentioned, I'm the founder and CEO of Swan ai. Um, it's a [00:01:00] startup. We're building the first autonomous business. Um, it's a business that is designed to scale with intelligence, not with headcount. Um, a business that is entire architecture is optimized around human AI collaboration rather than mm-hmm human to human coordination.

Um, briefly about my backstory. So this is not my first startup. I've built and scaled and exited the two startups before in the B2B world, both of them in the customer communication space. Um, and both of them were based on the old, uh, unicorn growth at all costs, playbook, where. You raise a lot of money before you even have initial signs of product market fit and you don't know who you're selling to, and then you grow to 30, 40 people before you even reach your first million dollars in revenue.

And then like every round becomes another story of how do we grow the vision that the potential and less focus about the metrics And that, [00:02:00] you know, two time journey really led me to believe that the startup model is broken. It hasn't been innovated on for the last, the last 15 years since Eric Rise have invented the MVP methodology in the Lean Startup book.

And so together with this revolution of AI agents, I finally came to the realization that it's time to reinvent. How do we scale a business in the model world where AI agents are part of the workspace? And you can finally scale with intelligence, not hit count and not capital. 

Mehmet: Right. You know, like, this is, this is a, a, a very, you know, rich introduction because you mentioned a lot of things.

Uh, I now, um, when you told people like you're building the company with just like three people, like what reactions did you get? Like, uh, what kind of, um, you know, things that might surprise you or maybe, and maybe [00:03:00] even of the things that you expected that people will, will say, okay. You know, we, we, we understand what he's trying to do.

Amos: Yeah. Um, so I was actually very surprised. So, because it's not my first startup in the, in the previous two companies, when we were in the early stage, I felt like there's something to hide that we're a small team. I always try to, um, create this sense of, um, of a bigger company. Basically, um, when we were five, I would say that, you know, yeah, we're, we're closing in on, on, on almost 10 people.

Um, we were, um, you know, 12 people say, yeah, we're almost 20. Right? I always try to give the, the customer or the investor the feeling that we're bigger than what we are. And now with Swan, we actually take pride of the fact that we're just three people and the reaction from the market is amazing. People, they look at us and they say, wow.

Just the three of you built that software and you're now talking with me. That's amazing. How can we be more efficient and more agile and more lean like you are? And [00:04:00] that changes the entire conversation dynamics because they're more attentive to what you're suggesting. And also. They're more supporting when you say that.

Sorry, I didn't have time to get that. We're just three people. So it, it created kinda like this weird dynamics where customers, they understand that we're building a different model. They aspire to also try to build something di similar to that. And now they're more understanding of when we have mishaps, when we're trying to follow that extreme new innovative model.

Mehmet: Great. Now let's little bit deep dive in, in, in, you know, what you built at Swans and you know, you call the Swans operating model. If I want you to describe to me almost like a day in the life, let's say, of you the three humans, right in the company and the roles of the AI agents also as well. So how that would look like for you.

Amos: So to take one step backwards [00:05:00] before that, um, I'll start with like our design philosophy around Sure. How we're thinking about human AI collaboration, and then we can take my day, like a, a single day in my life to actually see how it manifests into my routine. So our point of view is that we don't use AI agents to replace workers we would potentially hire.

So it's not like. We would've hired a support rep to do this job, so we could build an AI support rep agent, or we wanna develop pipeline, we would've hired an SDR, so we would, you know, build an ai SDR. That's not the the path we're taking. We think that this is not the right path to take. What we're doing is we're asking ourselves the question, how could we scale each employee within the company to their 100 x version of themselves, the 100 x engineer, the 100 x product, and in my case.

The 100 x growth revenue person, right? Um, and so if you look at my responsibilities, I'm in [00:06:00] charge in the company from everything from top of the funnel, generating demand to meeting with that demand and explaining about the product and trying to sell them the product, and then eventually onboarding them to the platform and maintaining them, um, and retaining them.

So what we did is gradually step by step every. Process in that funnel came into obstacles and challenges that came with us scaling and me having to drive more demand or meet with more customers, and every challenge like that unveiled in front of us. How could we unblock some of these challenges with AI while still.

Amos is in the center of the loop. So it's a human-centric design system. What we ask ourselves, it's what is it within the zone of genius of Amos? Mm-hmm. And what's outside of it. Okay. What's the zone of genius? It's where my passion and skills intersect to create disproportionate value [00:07:00] for the company.

And so if it's within my zone of genius, we want agents amplifying my routine, and if it's outside my zone of genius, we want agents. Automating that. So if you take my day to day, my routine, basically it revolves first of all a lot around storytelling. I love telling stories and I have a very vocal LinkedIn profiles.

Um, I share about the journey of building an autonomous business and we have millions of impressions on my posts. And so we don't want the AI to automate. Right, right. We keep wanna keep that as within my zone of genius. And so we built a system that the AI can monitor the engagements of the post and surface high quality leads around it.

And then the AI could help me reach out to the folks that went on our website and the AI could help me prepare for calls. So I don't need to spend time before calls. And then after the calls, automation of follow ups and CRM updates. And so what we did is we automated like the mundane tasks around my routine and we enabled me.[00:08:00] 

Through amplification to drive more value from the processes that I have passion for and good at. 

Mehmet: That's, you know, very eye-opening, I would say. Amazon, you mentioned, you know, about, you know, this balance between, you know, what you are best at and you mentioned like about storytelling, right? And you know, I, I can see the passion when it comes to GTM and I'm seeing a lot of mixed feelings, right?

Over there and recently not in the GTM space, like every single founder or leader in, in, uh, in the industry, whether they are building agents for, let's say HR or building agents for finance and so on. They're talking about keeping the human in the loop, and I think this is exactly also what you meant, what you're doing, like the, the human judgment that should be there.

So which part do you think really we can, you know, completely automate and we can know for sure [00:09:00] that people would not have any problem dealing with an agent When I'm talking here in the GTM space, of course. Uh, while always you think. The human would be required to have an input at least, or maybe a review to, to, to take the next action.

So what are like the boundaries here in your opinion? 

Amos: So that's a great question in my opinion, is that the boundaries are contextual to a specific company. Always. There isn't one answer that fits all. And the reason why is that if AI improves, um, it can start taking more and more jobs, you know, autonomously, and you compare the output.

And the value of that output to what? To the work that you could do within the company, right? And so what happens gradually is that you as a company should ask yourself, what is, is it that you're good at? What is it that you are [00:10:00] kind of like quote unquote, zone of genius? And even if you don't really have a zone of genius, you can act yourself.

What is it that, that our DNA that we want kinda like to maintain that gives us an advantage over competition? And I'll give you an example. If you're a go-to-market organization that is really good at cold calling and you develop that culture of cold calling like ZoomInfo for example, if folks that you're familiar with that company, it's an a public company and the, the moment that you, uh, sign up, uh, for request a demo, like five seconds afterwards to get a call from one of their reps because they have an amazing cold calling culture and.

Replacing that with AI would actually eat up their own competitive advantage, because if they're using AI to do that, then other companies could use the same AI to actually drive these same outcomes, and so they would lose their unique competitive advantage in their market. And so what they should ask themselves is.

How do we automate email and LinkedIn follow ups, but we amplify our reps [00:11:00] to just sit on top of the phone all day long and just call people and convert more and more and more. And what we'll try to use the AI is to, you know, try to leverage the human input and the human work to scale more and more and create more pipeline.

And then if you take a different, like a second organization. That organization don't have any, uh, calling culture, but it's really good at social selling and LinkedIn Swan for example. So we don't pick up the phone. Mm-hmm. Um, we're really good on LinkedIn. I have a very good social selling game. I write posts and then people interact with that post and I create conversations in LinkedIn.

That's how I generate a lot of pipeline. And so for us it would make sense to automate maybe the cold calling. Aspect, but to amplify our social selling game. And what we will do is like if I would use AI to replace me completely and to automate my post generation process and the way that I think about writing posts and how do I conduct in these nuanced conversations [00:12:00] in LinkedIn, then we would lose our competitive advantage in social selling.

And so I think that the answer to what should you replace and what should you amplify is highly dependent on what are you good at in comparison to your competitors that are trying to get to the same outcome. 

Mehmet: That's great. And I like, you know, this. You know, clear categorization of the company culture between cold calling or social selling.

Can I ask question? Maybe not related to what you're doing at Swan, but you know, there is a debate also like in this age of ai, in your opinion, and because maybe you deal also yourself with different type of.

You know, are you seeing kind of a conversions or diversions? I don't know which term would be the right one here. More, uh, in term of, you know, these sales methodology, is it like, you know, we always, in B2B we talk about like product-led growth, we talk about sales-led growth, we talk about founder-led [00:13:00] growth and in the age of ai, you know, now there are a lot of noise there.

And I think, um, and the reason I'm asking you also Amma, because you know this. Depending on your ICP, depending on, you know, who you're trying because you, you, you help your customers with prospecting, you help them build lists and, and all that. So for me, as a founder, hearing all this noise and different, you know, um, kind of different opinions and you are like, now with serial entrepreneurs, like serial founders, if you want to advise me as a founder, which methodology is the right for me, how I would pick up one.

Amos: Yeah. So, um, unfortunately I have another clever answer, like before that is not like directly answers your question, but, and, and it gives value, but that there's no shortcuts unfortunately. 

Mehmet: No problem. 

Amos: Sure. So, um. The biggest problem with the go-to market teams, entrepreneurs that are usually like first [00:14:00] timers, is that they think about how should we sell?

How should we grow? And they're thinking about themselves. They're thinking like, should we sell in, um, you know, should we send outreach or should we pick up the phone, or should we grow with the product? Or should we have a sales team? Like, how should we grow? How should we sell? But it's not about you.

It's about how should your buyers buy. It's not about you, and that's the biggest problem that people don't understand If you ask yourself. Like, how should my buyers optimally would buy my product then That's a very, it's a much easier answer. It's not a very easy ans question, but it's a much easier one.

And you know, if you look at my buyers, so we target, you know, startups, um, from C to series C. Um, that don't have a big rev ops team and go to market and they just, you know, they wanna deal with all that technical complexity that they have and just wanna use AI to remove all that technical complexity and just move 10 x faster.

Mm-hmm. And so I know that their buyer [00:15:00] journey would be very product driven because they just, they need to feel it. In their hands if the product actually is solving their problem. So what I wanna do is just push them into the product as fast as possible in their journey. But I know that for them it's easier for them to talk to someone before committing to an annual plan.

So I know in the buyer journey, I know that they want to talk to someone, so I need to identify the moments that they wanna talk to, to someone, and catch these hand raises the moment they need help. And so I'm not, I didn't use PLG or um, LinkedIn automation or any of these terminologies and playbooks. I just asked myself, what do my buyers ideally want and how could I support them into creating an optimal and personalized buyer journey that is a highly customized to my ICP?

And that is a much easier question to answer. 

Mehmet: And this is what I call it, like selling with empathy, actually like putting yourself in your [00:16:00] ICP shoes, right? And try to see like, how would these guys like to interact with us? Do they like to try the product? You know, by themselves? Do they want someone to walk them through and you know, probably maybe do a demo, whatever that is and yeah, like you need to define this.

Yeah. You want to say something? Exactly. 

Amos: So, uh, because I wanna give like the counter example. So this, my ICP is like, you know, seed to series C startups. And so that's how, like I realized that's their unique buying journey and optimal buying journey. But if I was selling, I was selling to big enterprises like CIO of, uh, you know, fortune 500 enterprise.

Of course their discovery process won't start by using the product. Right, right. They, first of all, they're not the user. They would not probably use the product, so they don't care about the product. They care about business outcomes. And so how do I, first of all, get them to interact with my brand in the context of solving and moving their business outcomes in the right [00:17:00] direction?

Right? That's the first thing they wanna know. And then if I could answer. That need to move a specific business outcome of a CIO. The next thing is to identify who they want to loop into the conversation. Who could take care of all these interesting things that they don't care about? Like how does the product work?

What, what about compliance? What about security, et cetera. And so your buyer journey starts with maybe the decision maker, but then tries to navigate within the organization and with different, different stakeholders. That are interested in different areas of your value proposition, and that's a totally different game.

This is called enterprise sales, right? But I didn't say enterprise sales until now. I was talking about the buyer journey, and that's where you should start from. And then when you mapped your buying journey, you can map accordingly. How should your growth model look like? 

Mehmet: Great. Now let me ask you something almost, um, regarding, you know, automating the outreach, right?

So, so this is what, what you, [00:18:00] what you help, uh, um, companies to do. Um. When we do these automations, when we do these, you know, GTM outreaches, like one of the things that we've, we've seen is like when you try to use ai, you know, for even simple like emails or like LinkedIn connection requests and all this, there are some situations where even if you try to personalize, but still it sound robotic, right?

Mm-hmm. So, you know. What, what is the right approach to do it here? Like I gotta be frank with you. Even if I know that I'm looking for something, and even if I know someone use ai, even if it's not a, your system, let's say even like a a a person, an SDR, maybe they wrote the message, but they use AI in the right way and they customize it.

Probably I would reply because at least it. They will show me that they did some [00:19:00] research before reaching out to me, right. To see if I, if, if I'm their ICP. Whereas like if I see this, what we call them, spray and pray messages, you know, and it's very obvious it's written by ai, probably are gonna ignore it.

So how we can make it like more human here, um, when, when, when you do the personalization for reach out. 

Amos: Yeah, so quick context about what do we do at SWAN is important because like automating outreach is, um, a very minor part and not always the case actually with our customers. So, um, for the audience here, what we do at Swan, our customers, they call Swan lovable.

For GTM, it's actually an AI go to market engineer. Something between developer and rev ops that works with sales marketing founders. To turn any go-to market process into an agentic workflow in seconds, you know, from prompt to pipeline so you can really scale, demand and revenue with intelligence, not [00:20:00] headcount.

And so for these companies, their biggest challenge is actually that their tech stack and technical complexity, and they have a so like idea and they want to perform different outreach campaigns or different types of uh, uh, intense signals or inbound flows or outbound flows or pipelines. Processes, whatever.

They have these ideas, but it takes them so much time to build out these AI based workflows. So they just, they don't do it. And then Swan comes into the picture, you just chat with it, you tell what you want. And Swan is deeply integrated your tech stack and can build out these, um, automations and agents basically across your entire tech stack, so you could move much faster and use the most advanced technology right now in your favor.

And in that context, part of what we do, we help. Companies run outreach. Some of them run fully automated, some of them run human assisted, where humans are actually, you know, on each step of the sequence and they review each email or each LinkedIn message and you can choose how do you wanna run that. But to your question about [00:21:00] personalization at scale, you know, kind the secrets to running that successfully.

So. There are a couple of, um, best practices that about how to do that well to actually drive results and generate more pipeline. And the first one is that no one cares about personalization. They care about relevancy. So relevancy beats personalization. Personalization could be a met meant, I saw that you're running a podcast.

Would you like to buy my GPUs for, to run your AI better? That's personalized, but it's not relevant. Right. But if I was able to say, Hey, I saw that you're actually released, you just released a new AI product right now that is using kind of the latest AI models. I thought maybe you might consider, um, running it on a different GPU infrastructure that is relevancy.

Right. And so, uh, to actually drive results and outreach, you need to think about relevancy and not personalization. That's tip number one. [00:22:00] Now second tip is that AI is as good as the context it has. And that a lot of people don't really understand today is that if you just tell Chad GPT write a personalized email to someone and it doesn't have any context about what you do, your value proposition, um, how do you handle different personas, uh, different outreach angles based on different signals and events that are happening within that business, et cetera, then the AI won't really be able to move the needle and it will sound generic and you'll be able to actually feel like, yeah, it's not really, you know, talking to me because.

Doesn't make really sense. So, so the more context you give it, the better the AI works and that's why there's a human in the loop. And we talked previously, Matt, Matt, about the importance of the human input in the GTM organization. So context engineering, that's the term actually that is emerging right now.

So context engineering is becoming, uh, the craft of go to market teams and it's [00:23:00] becoming the most important input of humans in the process is context engineering. How do we. Inject our knowledge of our ICP of outreach strategies of different approaches and angles into the ai. So it is able to really create relevancy at scale in a consistent and accurate way.

Mehmet: That's, that's, you know, the, the, the, the exact way where I'm thinking also today of, of doing things. Um, you know, when, when I talk to, to. Founders, um, you know, and the term is, is maturing also as well. I think Amos and, um, the GTM movement is also shaping what we used. It's still called actually product market fit right now.

But tell me how. The go-to market. One is driven by agents also. I mean, you have the agents on your side how this can help. Like can, can I [00:24:00] get early signals, which would tell me like, yeah, I have kind of a product market fit here, so I need to double down Maybe. On the efforts I'm doing, or maybe it'll give me the science, like, Hey, you need to go and pivot because, you know, we're not seeing the, the responses that we're waiting for.

Maybe we're not getting enough inbounds from, from the campaigns we're trying to do. So how. You know, this age of AgTech, ai, autonomous ai, whatever you want to call it, is reshaping. And, and let's say, um, fastening also like, uh, speeding up the product market fit in your opinion. 

Amos: Mm-hmm. Yeah. And, um, I'll double down on the last sentence because it encapsulates the, the, the important aspects of what I'm about to say.

It's, it's not about reshaping, it's about fastening, it's about speeding up. Okay. It's not, it's not about reshaping because nothing has changed. I think people think that, you know, agents come and they, we need to, to forget about everything, 

Mehmet: do magic 

Amos: and, [00:25:00] and yeah. And they, they, they will just solve all of our problems.

But no, you know, GTM foundations. Are as equally as important today, if not more than ever. Basically if we talked in the previous section about context engineering. Context engineering in go-to market is highly dependent on your GTM foundations. Um, but to your question, what does it look like to use agents in an autonomous business to fasten to, uh, to speed up the process of getting to product market fit?

And, uh, what we took, first of all is a page out of Brian's Halligan playbook, the founder of HubSpot and c former CEO, uh, which treated pre ai, which treated go-to market like a product. It said, it said GTM is like a product, and you need to think of it like that. And product is a method, you know, developing product, like developing GTM as a methodology of iteration, experimentation, getting to something that works and then [00:26:00] trying to scale it, um, so that you can input small input and create and get massive output.

And so that's how we look at go to market as well. And swan. So we start by. Fast iterations and experimentation on a specific hypothesis. We think that we could generate pipeline through my social LinkedIn activity. Um, we had an hypothesis for that because I was starting, you know, to write posts and I thought, and we were getting engagement, so we thought maybe we could double down on that and just generate pipeline out of this channel hypothesis.

Let's start iterating and experiment on this. So agents allow us to experiment faster. Why? I use agents, as, you know, an extension of my thought process on how to write these posts. So I can write posts, I can spend less time on writing these posts and get, actually, get a better quality, uh, because of that.

Then when we release a post, I can use agents to monitor the engagement so I can actually see who's engaging with my post and [00:27:00] qualify them. Understand how many of them are in my ICP. I can understand how many of them actually turned into qualified pipeline, right? So I can actually verify that this motion is working.

Okay. We've experimented, we've iterated, we realized that this channel is starting to create pipeline. How do we scale it? And now what we can start doing is identifying scaling bottlenecks within that process and, um, dividing the labor between the humans and the ai at each point of that challenge. What we're asking ourselves, we're not always asking, how should we automate this with ai, this is the wrong question.

We start by asking ourselves. Is this something the human should do? That's the first question we ask ourselves. Hmm. And we, and some of the answers, you know, if I'm thinking about what to write and how to write and how to tell the story of a post, I ask myself, is this something that the human should do?

Yes. 100% yes. Because if I ask the question, should [00:28:00] AI automate it? I could, yeah, it could automate that, right? But I don't want it to automate that. But then if you look at monitoring the engagement of my posts and, you know, analyzing 1000 likes of a post. I don't think a human should do that, right? We could automate that with ai.

So amazing. Let's do that. And that's how we're scaling a specific channel by asking the questions. Should a human do that? No. Can the AI do that? Yes. Amazing. Let's deploy an agent there. And that's how you can scale it. So with little input, you get more and more output as you scale that methodology or channel.

And that's how you use agents and go to market. By speeding up the process of ex hypothesizing, experimentation, iteration, and scaling. 

Mehmet: Let me ask you as a follow up question, which is a question I prepared, but now it came in the in the right time. Is executing fast and experimenting fast [00:29:00] and failing fast.

Becoming more and more kind of the new mode for startups in European ammo. Because if you think about it, like everyone has access to the models, everyone has access, you know, to, to a lot of things. I know like the data is, is for sure 1, 1, 1 aspect of it, like especially if you have kind of a property data which you trained your.

LLM models on it. I don't want to go into the technical details, but looks to me that also being agile and having this willingness to execute fast, the speed, do you think it's becoming more and more kind of a mode also as well, uh, in the age of AI agents? 

Amos: So the answer is yes and no. Okay. Um, and the reason why it's Yes, it is a mode.

Is it becoming more of a mode? No, just because it was always the mo in my opinion. It was always the only thing that a startup could do. In front of [00:30:00] the incumbents. So Google could out market you, they could outsell you, they could out engineer you, they could out everything you, you don't have any structural advantage over Google or open AI or Microsoft, whatever, besides your decision making velocity, you can make more decisions at a given day than Google because you have less bureaucracy, less constraints, less people to coordinate.

Less meetings and less approval chains so you can make more decisions than Google. And so if there's an idea that the only thing that is standing between you and realizing that that is an idea that you should double down is just decisions. In executing on these decisions. So if you can increase the speed of your team making those decisions to get to that idea, then you can realize something before Google and you can execute into the right direction.

It's [00:31:00] faster than Google. That's what you can do. And the only reason why companies like Google acquire startups is because they know something that Google don't know because they're executing in the right direction faster than Google can execute themself. Right. And. Basically when people ask me about SWAN and as an autonomous business, about like, is this just about not hiring people?

Are you just trying to prove the world that you can be like hyper lean and efficient team? And my answer is no. It's not about that. What we're trying to show is that the way for a startup to scale. It's by creating leverage and in increasing the decision making velocity of the business. And the smaller the team, the smaller the ship to navigate, the faster the decisions are being made because you have less approval chains and autonomous business.

It's not because we're using AI agents, it's because we're optimizing for autonomy. Autonomy of humans and autonomy of agents collaborating together. The more autonomy you have, the [00:32:00] less approval change you have, the faster decisions you can make, the more leverage you create, the faster you scale. 

Mehmet: This is interesting and.

You know, this brings me to, to ask you about, you know, the, the growth and, you know, maybe fundraising. Of course there is a lot of talk. Um, you know, Sam Altman mentioned it and a couple of other people mentioned about the one, one person, billion dollar company and all this. Are you seeing the ecosystem is ready When I say the ecosystem like investors, VCs, you know, uh, and everyone in, in between for this new.

Um, generation of companies which are run by, as you mentioned, like very um. Augmented, uh, team that they are trying to leverage, you know, the, the power of the technology to their side. And actually may, maybe it's a loaded question, sorry for [00:33:00] that, but would would still, you know, be, um, a requirement to grow the business re you know, requiring maybe some funds in the future?

Like maybe the first bootstrap will become like the default more. Are you seeing like, kind of shifts to towards that? 

Amos: Yeah, so. First thing is that I don't really like the one person unicorn terminology. The reason why is because, you know, the, the unicorn is a mythical creature, right? It doesn't exist in the world, and you try to kind of create an idealistic, you know, creature that is not really feasible in some sense, right?

There's something about that. And so you take the one person unicorn. It doesn't sound like a model that you can replicate really, it sounds like. Well, there might be. You know, one person who cracked that and will just do that once in a generation, they will be the one person unicorn. Right? And that's not what we're trying to do at Swan.

What we're doing is we're building the playbook of how, you know, startups would scale in the next [00:34:00] five years. And we think that this is the model that should be replicated. And so we're not building a unicorn here. What we're building, that's actually the origin of the name of the company. We're building a swan and a swan.

Unlike a unicorn, it's a real creature. And it starts as an ugly duckling, right? And it doesn't have any TechCrunch news. It's not famous around the world, huge funding rounds or anything like that. But, um, through rigor and hard work, uh, and grinding, it starts, becomes this beautiful, aerodynamic, elegant creature that is, it's, it's humble in some sense because it's just white, but it has that beautiful glow around it.

People look up to it, right? It's not like a peacock, right? It's, it's, it's different than that. It's a swan in that sense. And so it's an important analogy here, ma, because we're not building a unicorn, but it doesn't mean that we're against fundraising or something like that. What we want to show is that the dynamics.

A venture capital [00:35:00] backed business could actually look very differently in a world where you could scale with intelligence, not necessarily with headcount and capital. And so we're, we actually get a lot of, um, inquiries from investors, and we're not saying no just because we don't, we're not intended of raising money.

It's not about bootstrapping. And we actually see a huge interest from investors. To rethink how are they thinking about investments in an AI native world, right? How are they thinking about early stage investments where startups could take small amounts of capital and grow massively with that? And what we wanna show the world actually is not that you don't need capital, this is how bootstrap businesses should grow.

What we wanna show is that how you can rely on very little capital to build massive companies that generate massive returns and massive impact on the market. 

Mehmet: You know, like this is, this is a great, I would say, hmm. Role [00:36:00] model for, uh, other entrepreneurs also as well. I would say almost. And you know, I, I like, you know, your way of being authentic and this is, it's no surprise because you mentioned you have, you know, the, the, the, the social, uh, media power behind you also as well.

And you, you can do this without authenticity. You can't do this without, you know, real support from, from your audience, which is. It's in my opinion, another mode also as well, like building this, uh, trustworthiness, I would say with uh, with people around you is very important. And this is where you know my, it's not a question more like a my thought.

And I would love to hear your thought now because you are like able to build a full fledged company with less people, the culture and leadership. Stay, you know, kind of humble and nimble and doesn't change a lot because you know when. And always discuss with [00:37:00] multiple, um, guests on, on the show about maybe the culture.

And you know how there is the secret number? Like some people, they say it's 50. Some people they say it's a hundred. Some people they say it's 200. So once you cross this number, things start to be break up a little bit. And it's, it's nature because the more you have people on your team, so you have like first line managers and now you have maybe like, even like two lines of managers in between and things start to change.

So. Is this, in your opinion also like a way where we can keep the, the culture I can see very clearly amo, you have this culture of authenticity, the culture of the customer first approach. So is leadership becoming, I'm not say easy, but it's all about, you know, this being close to your customer and you know, having a small team of people you trust around you.

Yeah, I'm just, 

Amos: I 

Mehmet: would thinking out loud. 

Amos: No, no, I, I totally get the question. And I, I would argue that, um, [00:38:00] in that context, leadership is harder, not easier. I would explain why. Um, so in my opinion, in my opinion, leadership is given not taken. Right, and when you take leadership, then you need to create power structures around it to ensure that your leadership persists, even though there is mis compliance from.

You know, your subjects, let's call it like that. Um, we're trying to look at it from an hierarchy perspective, and it's important to talk about a hierarchy structure here, because that's what an regular org chart creates, right? It creates hierarchy. Hierarchy is the equivalent of, you know. A clear power, dynamic structure, and a corporate right?

And so it means that you are obliged to perform in this specific manner if your manager tells you so, because you're under that manager in the org chart and the manager can fire [00:39:00] you, right? 'cause they're responsible for you and they're leading you. And the reason why I think it's harder in an autonomous business where, um, autonomous business thrives to have as flat as possible, uh, org chart as it can.

It's not like a dogma. We're not religious about it, but you strive to make it, uh, flat because flat is the equivalent of autonomy in that sense, right? The higher autonomy, the more flat the organization is, and the reason why it's. More difficult to lead in a flat organization is because no one is really owed to listen to you.

Right? They, they, they, if they don't want to, they won't because they can do whatever they want. They have full autonomy. And so, you know, I feel it a lot of times when I, of something that I think will be very important for the business and for example, um. I think that our product should go into a specific direction.

So context engineering is emerging, and I can see that, um, from a visionary [00:40:00] perspective, because I'm at the CEO, um, position and I, I look far ahead. I, I see this is an a direction we should double down on, so I wanna kind of direct the product to invest more in that area. Right. But the product doesn't listen to what I say.

Right? They don't, right. They don't have like, uh, they're not compelled to listen to what I say, so I need to convince them to listen to me. Right? And they need to feel from the other side that if I come to tell them something, they should listen to me because it would create a better product for them, right?

We want to create that aligned incentives in that sense. And so it's really hard to create this culture where. Other people that are doing jobs that are not really directly related to what you do in your day-to-day. They want to get you involved in their business and they wanna listen to your opinions.

And when you tell them something, you, they, they would listen to you, but sometimes you need to accept the fact that they won't listen to you and they will do something else. And you need to trust them that because you give them the keys to [00:41:00] move the goal that they're leading. Right? So leadership in an autonomous business is much harder and is much more driven by.

Um, charisma and organizational understanding and your skills and your ability to communicate and to convey value and to understand the other side that you're talking to and to kinda like tie them into your mission. Um, and it's actually much harder to drive changes within an autonomous business because of that.

Mehmet: I love that. Uh, you know, and, and yeah, indeed, because you have more responsibility. I, I believe also, I mean, you, at least you feel it, you know? And it is there. It's not like just a feeling. I agree with you a hundred percent as we are coming closer and, and really I'm enjoying the conversation. Like what's exciting you about what you're doing in the future with Swan, like anything you might share with us today.

Amos: Yeah, so the reason why we built Swan. Is like [00:42:00] I would call it, uh, we're building an institute to research, um, human AI collaboration. Okay. That's what we're basically obsessing about. We're obsessing about human AI collaboration and the more that the, this company will grow, I, I hope that the more we'll be able to progress, um, you know, human's knowledge about human AI collaboration within the context.

Of B2B businesses. Let's talk. Let's try to confine it into a specific niche right now. And I think we will grow beyond that niche, the more we will grow. But what we're obsessing about right now is how do you build a business that serves other businesses, and what does it mean to have human AI collaboration in an environment like that from a perspective.

Our perspective from our customer's perspective, from go-to market perspective, from r and d perspective, from support perspective, from leadership perspective, from every perspective of the business. What we're obsessing about is defining the rails for people to be able to drive [00:43:00] better outcomes from human AI collaboration.

And that's what we're obsessing about. And we have a lot of. Cool things that we're building and sharing with the world around that specific subject. And I hope that the more we grow, the more light we could share around the world without specific story in mind. 

Mehmet: That's great. And so for people to stay tuned where they can find more Amos, more about you and get one.

Amos: So, uh, I have my LinkedIn, right, Amos Bar Joseph. Uh, and if you wanna, you know, follow me there, that would be great. I share a lot of, uh, learnings about my journey. Then I have a newsletter, um, it's called The Big Shift, um, of Swan ai. So I share like a back seat story. Um, it's actually seeing how we're building our autonomous business and the wins and losses and frameworks.

And I'm sharing a lot. What about my thinking? Um, there. And finally, if you had any questions that you wanna ask me. So I have a digital clone basically that you, it's not like me, it's not as good as me, but it has [00:44:00] access to all of my playbooks and posts and everything that I've wrote. So, um, it's called autonomous and it's available in the GPT store in chat, GPT.

And um, we are gonna drop the links after the show me, I'm sure. So if you wanna ask a couple of questions there, you're welcome. 

Mehmet: Yeah, of course. We are gonna drop the, the links there. Thank you for managing this Amos. Um, you know, as I said, I really enjoyed the conversation, what, what you're trying to do.

Um, I can see it's not about building something cool. I can see it's about building something meaningful and, um, you know, and this is my take for today that really. Um, and this is for, you know, some of the folks that you know, they are, I can understand, you know, the skepticism when, when you tell them someone is building, especially something similar to what you do with AI agents.

You see, oh no, like, you know, like, that doesn't work, blah, blah, blah. So, you know, like what you just. [00:45:00] You know, showed us today is, it's the opposite actually. It's about keeping, you know, this human touch, um, you know, about leveraging, you know, the, the superpower Actually, we, we, we are super powerful and we have now the tool to be like double super powerful and, and you know, I'm happy that you shared, you know, what you, what you're doing ammo.

Yeah, again, uh, great to talk to you today and as you said, as AMO mentioned, all the links, uh, for the website, for his LinkedIn profile, for the newsletter, for his CLO Chad GPT store. There will be, uh, all, um, available. In the show notes, if you're listening on your favorite podcasting app, if you're watching this on YouTube, they will be in the YouTube description.

And this is how I had my episodes. This is for the audience. If you just tuned in, thank you for passing by if it's their your first time. I hope you enjoyed it. If you did, so give me a favor, subscribe and share it with your friends and colleagues. And if you are one of the people who are like loyal [00:46:00] to the show, they keep coming again and again.

They keep sending me their feedback. Questions wishlist. Thank you for doing so. I read all these emails and notes, and if you know you are, uh, again, trying to reach out for maybe mentioning a guest you wish to see on the, on the show, please reach out to me. So, uh, 2026. We have a good lineup so far, but of course, you know, if you have someone on your mind, happy to discuss this also as well.

And finally, big, big thank you because you keep pushing the show. In 2026 to the top 200 Apple Podcast charts across multiple countries. So we, we, we keep roaming in different countries all the time, and this is a trend since last year, 2025. And thank you for, you know, tuning in and listening because without you, this show couldn't sustain that long.

So thank you very much. And that is always, stay tuned for a new episode very soon. Thank you. [00:47:00] Bye.