GET ON THE FIRST PAGE OF GOOGLE!

Increase traffic by 1200%!

AI Automate Your Business And SEO Martin Atrin

AI Automate Your Business And SEO Martin Atrin

Live from Chiang Mai with Martin Atrin

Yannis: We’re live! This is Yannis Divramis — hashtag divramis, divramis.com — we’re on every social media. We’re here not at Shenzhen SEO but in Chiang Mai, with a friend who is organizing local meetups. He’s very specialized in AI; he’s a programmer, he comes from Slovakia, but he lives here in Chiang Mai as a digital entrepreneur and digital nomad. He’s going to present himself to us and to all of you, and you can also ask questions about AI and AI automation. Because many people are talking about AI, but they don’t make any money from AI — and this guy really got the point. We’re going to the hackathon on Saturday together, because you invited me there. And the reason we’re doing this special interview with him, live — and it will be on every social media — is because he knows a lot about AI: how it can automate your business 10x, 20x or even more. This guy is called Martin Atrin, and we have him here with us today.

Martin: Hello guys, nice to meet you. I can see a lot of Greek. Are there a lot of Greeks? Yeah — let’s speak Slovak. If I speak Slovak, maybe some of you will understand, because when I was in Bratislava we had a lot of Greek students there.

Who Is Martin Atrin

Martin: So, my name is Martin Atrin, and I basically specialize in the deployment of AI — architecting different solutions. I’m on the path to becoming a professional — not basic — Anthropic certified architect. I’m in partnership with OpenAI, and I’m an ambassador for Northern Thailand for Z.ai — for the GLM models, if you know about those. And I pretty much cover everything from software to hardware, primarily as a consultation, and then development and deployment as well.

I’m currently part of two startup projects with two companies. One of them is called Kybernesis — kybernesis.ai. That one is specifically focused on companies that are more on the medium-to-enterprise side: companies that really care about data privacy, that really care about durable processes, that want everything to be bulletproof. The second company is called Tilbrick, which we are in the process of finalizing; I think the first version of the Tilbrick portal will come online sometime next week. That one is aimed more towards self-hosting clientele — people who are primarily doing things themselves — so it helps them work with AI on a much higher level, and it’s primarily based on subscriptions, whereas Kybernesis is aimed at customers who prefer API usage. So, two very distinct approaches to various harnesses. That’s pretty much it.

How Martin Got into AI

Yannis: So the first question is: how did you get into AI, and why?

Martin: I was always tech-savvy, and I was always around technology — software, hardware, everything around that — but never professionally. I used to own five e-commerce stores, and those five stores were doing quite well. I sold them, cashed out, took a five-year-long holiday, and I just kept learning. And then AI happened — suddenly learning became easier, so I could accelerate my process way faster, and I’ve gotten myself to this point. Now I can confidently say that from start to finish — even if you need to pre-train a model — I’m quite capable of executing everything.

And I got into AI because it’s freaking awesome. Generally speaking, if you abstract away the “oh, it’s living, it’s not living” — it’s a tool, a tool like anything else. Nobody really questions email; everybody questions AI. Some people don’t know how email technology works; they take it for granted. It’s very simplistic, and the same can be said for AI: it’s actually very simple and straightforward. But what it unlocks is basically the distilled knowledge of the internet, and when you pair it with the correct tools, it can help you leverage the knowledge of the entire internet towards whatever your goal is. That’s actually how I got into it.

I made it my profession to help businesses realize: “Oh, we might need AI for this, because there is a person doing something repetitive that an AI can do.” And during the auditing phase and the consultation, sometimes we come to the conclusion that they were too fast to put AI somewhere where AI is absolutely not needed — and it actually produces better results not to have AI in that process. So it’s very varied, but at the core of everything it’s still just basic software development, and I like that. I like that very much.

Yannis: Nice. It’s a very good approach.

Can AI Make You Money?

Yannis: So, Martin, you were in e-commerce, like many people still are. Many people are trapped in e-commerce and they want to get into AI like you. What did you see coming — that AI can make money for businesses? What was the thing that made you say: this is different, someone can make money with AI?

Martin: That’s a good question, to be honest. But in complete honesty, AI itself won’t make you money. AI always has to be paired with domain expertise. Let’s say you have e-commerce in a specific segment — the one I earned the most money in was cosmetics. So you have an e-commerce store in cosmetics: you probably know something about those cosmetic products, so you have domain expertise in terms of how to sell what you are selling, and how long the logistics cycle is — from ordering from your manufacturer, through interim warehouses, all the way to the last-mile delivery to a customer. Now, when you take all of this, you need to break it down into very small logical loops, and the moment you do that, you will start to realize where on your process map AI can help you save.

And this is the first important segment. It’s not about making money; with that first segment you already have a business — let’s learn how to save money, how to save effort, how to save time.

Second step — we hope we are already saving money: how can we scale certain professions within our company in order to earn more? How do we approach marketing in the era of AI? What information are we scraping? How are we working with the data we actually get? AI can help you all the way. Something that used to be a profession — and, to be quite frank, still is — like a Facebook ads guru or a TikTok ads guru: now you can replace that person at the medium level. I’m not saying senior, or the best there is, but at the medium level you can hire that person with a subscription to AI. So suddenly, if you were doing just some ads, you can start making ads on every single platform.

And there is a new trend that I’m trying to start. Are you familiar with A/B testing? In the era of AI, it’s A–Z testing. Because instead of trying two campaigns, you can let AI programmatically make 15, 20, 25, 30 campaigns and let it evaluate them in real time. So you start with a really wide base, and then you narrow it down to the best copies, the best ads, the things that really stick. That doesn’t discount that you should always make quality content as the basis of your ads, as the basis of everything you do — but you can stretch it so much further using AI.

And last but not least is making money with the actual AI — and that’s what I do: actually getting AI into companies. But as with every profession, when there is hype, there is saturation. There are a lot of people doing it, and there is also a lot of bad taste in the mouth, especially when it comes to businesses — because some of these businesses have already been approached by a super-motivated 17-year-old who told them everything was going to be amazing, and they paid him some money, because the young generation knows how to do these things — but in the end he couldn’t deliver. So when you approach them later, even with the knowledge, you are going to have a much harder time. This is unfortunately prevalent in the sector, and that’s why pretty much all the companies that actually mean something are grabbing the tools that are official, that are supported, that have certifications, and they’re going for things that are a little more stable.

So be wary. If you want to start making money with AI, I would focus first on whatever you are doing now: how to make it more efficient, how to scale it to reach more audiences. And how to actually make money with AI by replicating the same process for other people should be the last of your concerns.

Established Businesses: Start with a Process Map

Yannis: Yeah, Martin, you mentioned a lot of things right now. There are businesses that are already established, and there are people who want to start their own business and are listening to your message today. So let me split the question: how can you help an established business improve its profitability with AI, and how can you help someone who wants to start his or her own business with AI?

Martin: Let’s start with the first leg of the question. When it comes to established businesses, the number one thing you have to take into consideration is that the business has to have a process map. Because AI can do so many things, it is impossible to do it well if you don’t have a scope: you need to know what you are doing and what the deliverable is. If I want to deliver a better marketing tool, then I need to know how marketing works in your company. That’s why every serious company that wants help with AI should keep in mind that the first thing they need to do is have themselves audited by an AI expert who will tell them: this is what you’re doing, this is how you’re doing it right now, and this is what can be better with AI. That’s the number one thing. Without that, you are just shooting and praying that something sticks and something helps.

The AI Audit

Yannis: Can you elaborate more — sorry to interrupt you — on the AI audit? Because you mentioned something very important. So the first step for someone who comes to you is an audit?

Martin: Yeah. We usually sit down, because I need to scope: what’s the size of the company, what are we going to focus on — is it sales, marketing, logistics, price offers? I literally had that conversation yesterday with a company that manufactures robots — actual robotic arms and everything. This company has such a complex process even for establishing a price offer, and this is something they need help with, and they need AI. Because to go from “Oh, hello, I would like to purchase five robots for the manufacturing line” to them actually delivering — or even signing the contract — takes two months, because they have to build 3D models of how it will sit in the space, etc. And AI actually helps a lot with this particular thing.

So the audit is scoped depending on the company size and depending on the target. During the audit process we basically drill down with the employees: their day-to-day, their work, what they are doing, how they’re doing it, which programs they’re using when they’re doing it. And this is the Kybernesis part — with the company Kybernesis we drill them down to the very basics, and we ask them how they actually want to interact with AI. Because everybody is trying to sell you their agent — you have Hermes, OpenClaw, you have all of these agents, and everybody says “Oh, take this agent, it’s the best agent” — but in reality you won’t even use it. Especially if the company is medium-sized, with 15–20 employees in a team: they are not going to learn a new tool just because you said so. They already communicate in Microsoft Teams, or somewhere else. So why would you try to force a new thing on them that just creates friction? So again, part of the audit is to actually investigate what they will use, and what the chances are.

Starting from Zero: Two Ways to Make Money with AI

Martin: We can go to the second one, which we call greenfield: we’re starting green. Okay — I have no company now; how do I make money with AI? There are two basic principles that are the easiest to adopt.

Principle number one goes back to the very, very basics of business: find people who don’t know how to use AI and teach them how to use AI. It’s a service: you basically come to them, sit down with them, install the ChatGPT or Claude desktop apps, set up a couple of schedules, and get paid. The running rate is roughly fifty to a hundred dollars an hour, and you’re just teaching people how to use it. It’s very simple. The same thing applies as it applied when websites were new: if you wanted to start making money with websites, you went to a business that didn’t have a website and sold them a website. Really simple.

The second, more advanced way is that you have some knowledge about something — about a specific subject. And I’m going to say something that’s very controversial, because this is a message you hear everywhere right now: “SaaS is dead. SaaS is finished. Nobody needs any SaaS.” You hear this everywhere. But at the end of the day, if you are using, let’s say, Claude or ChatGPT in a company setting, what is the first thing you do? You connect it to your HubSpot, you connect it to your Attio, you connect it to your Salesforce — you connect it to something that already has a data structure in a certain domain, in a certain vertical.

Let’s take the example of the e-commerce business. I have an e-commerce business in cosmetics. I don’t want to do it anymore; I want to close it down. I don’t want to use AI to make it better — I just don’t want to do e-commerce. What am I going to do? I know how the entire logistics process works. Therefore, my SaaS product is basically that entire process, step by step — and that process, with its app connectivity to ChatGPT and Claude, is your product. So SaaS is 100% not dead. Because for AI to perform at the maximum possible level, you need a frontier model, a specialized model, and tools. And tools are something you just can’t skip: every model uses them, every business needs them. So if you make a tool that AI can leverage, you are winning, because people will keep paying you five bucks, ten bucks, twenty bucks a month, depending on what your expertise is.

And at the end of the day, what is HubSpot? What is Salesforce? It’s just a massive database. It has rows: this row says the customer name, the customer address, something; this row says when an opportunity is there — the name of the opportunity, the size of the opportunity, when to catch up with that person. It’s just data. So figure out what you’re good at, build it into literally an Excel sheet, let your AI rework it into a product, and create adapters to AI. That’s how you can get started from nothing — which is 20 bucks — and actually make a business for yourself.

Where to Find Martin

Yannis: Martin, every time you explain something, you open new windows, and I don’t have the time to elaborate on everything. You can book a consultation with Martin Atrin. Where can people find you?

Martin: I do have a website, martinatrin.com. You can find my contact for WhatsApp etc. there, and on Instagram, martin.atrin. And if you can drop it — I think it’s in the title of this live stream.

Yannis: Yeah, it’s also in the title.

How AI Can Scale an SEO Agency

Yannis: And we have time for another question?

Martin: Yeah, sure.

Yannis: Let’s do one more question. Let’s say we have a business in the SEO industry — because we are an SEO channel — and let’s say we’re making fifty thousand dollars per month right now, fifty thousand euros. How can this business come to you, and how can you help it scale? Do you understand the question?

Martin: Book a consultation with me! Book a consultation with me first. But it depends on the level of AI utilization that’s currently there, because I assume most of the SEO agencies that are still alive today use AI to a certain degree. So first I definitely need to have an intro call, and I need to look at the processes. I need them to explain to me the rationale behind how they are doing things right now. If there is no AI, we can automate certain steps; if there is AI, we can look at how to do things better, how to organize them better — and, most importantly for SEO, how to version it, how to backtrack it, and how to test continuously and meticulously. Because if you want to stand out in the SEO game, it’s very much about single words, so in order to track everything as granularly as possible, there are a lot of processes that need to be made and then automated — so the SEO workers and the SEO agency owner don’t have to do it all by themselves, and they can focus more on the current trends and on their predictions for the future.

Because if there is one thing AI literally cannot do, it’s predict a trend. It cannot predict anything. The way AI works is like this: you have a bell curve of responses, and AI always chooses something in the middle. Do you want to see a magic trick? Ask your AI for a random number between 1 and 30 — I don’t care which AI you’re using — and in five seconds I’m going to tell you the number the AI will give you. It’s 17. And if it’s not 17, it’s 23. And if you ask it for another number, it will be either 17 or 23, because AI always gives the most likely answer from the middle. Therefore it is very hard to get it to predict something, to do something novel, something different. So try thinking about this for the future: use and leverage AI for everything that can be automated, for all the data abstraction you need to do, and for reasoning on top of that data. And don’t let it predict — that’s still your job, that’s still your expertise. It can churn through a lot of things, but…

Yannis: Martin, this is crazy. It always comes out 17 or 23 — we have to try it. But now we understand how AI works; Martin is an AI engineer and knows better how AI works.

AI Five Years from Now

Yannis: One final question? Do we have time, or do we have to go?

Martin: Yeah, go for it.

Yannis: How will AI be five years from now? Can you make a prediction?

Martin: Do you want a technical answer or a social answer?

Yannis: Social answer.

Martin: Okay, so, the social answer: it will work better. It will just work better. It will be better at capturing the intent of what you want to do with it; it may be able to predict what you want depending on your past behavior. But I think that’s pretty much it — with the transformer architecture that all of the LLMs currently use, there are certain limits. It will be faster, it will be cheaper. In my personal opinion, it definitely won’t be super expensive, as they’re trying to paint it, with Anthropic and OpenAI. It will most likely live on your device, because small models are catching up every day, and for the general population the thing that lives on your phone will be good enough.

And as a business, you will need to adapt to this new reality: people are not really searching for things — agents are searching for things. People are not really buying things, because instead of a list — the page of results you would normally see on Amazon for physical products — AI will distill it down to the top two that match the criteria. This is the reality you need to work with as a business, and it’s going to be an absolute bloodbath, business-wise.

AI-wise, I don’t even know if companies like OpenAI and Anthropic will survive the next five years, because the number of new players that can do the same thing — or very close to it — 20, 30, 100 times cheaper is increasing every day, and this will eventually become problematic for the big companies: to justify the cost of research, to justify everything else, and the entire company structure and how it works. So who knows, honestly. It will definitely be faster, it will definitely be better, but there is a certain limit. We had a meetup yesterday where I gave a full talk about the things that actually excite me in AI — and that’s a new architecture: a model that only directs me to a different model, rather than trying to give you an answer itself. All of these things together make for a really weird prediction about the next five years. I don’t know if I can give a concrete answer to that.

Should You Start an AI Business in 2026?

Yannis: Martin: start an AI business now, in 2026, or not start, because of the high competition that’s looming?

Martin: It depends on your budget, in my opinion — really. If your budget is next to zero, it’s a good learning experience that will help you in a job or a future business, regardless. If you have decent money to invest, or you already have decent revenue that you can sacrifice towards the future, then you will probably win, because you’ll be able to accelerate way faster than anyone else. The key differentiating factor, especially in B2B: do not forget the basics of business. Relationships really matter in B2B; no amount of AI will scratch that. Try to orient yourself towards an expertise you already have, and just use AI as leverage. That’s probably my number one tip for 2026.

Thank You, Martin

Yannis: Okay, we have to thank Martin Atrin. This is me, Yannis Divramis, live from Chiang Mai, before the Chiang Mai SEO Conference 2026, with the specialist who is organizing local meetups and sharing his knowledge — the most knowledgeable AI professional I’ve found here in Chiang Mai. Where can they find you?

Martin: At martinatrin.com, yeah. Also on Instagram, Martin Atrin, and on YouTube — I do have a YouTube channel as well. Give me a second… my meetups are actually there. The meetups are there, and they’re very fruitful.

Yannis: You can find them on the YouTube channel, under the name Martin Atrin. Thank you very much for being here with us today. Follow me on TikTok — hashtag divramis — or at divramis.com, or call me on WhatsApp or Viber, because we’re in Chiang Mai for SEO services and the latest trends in SEO. Yannis Divramis — hello to everybody.

Book a Discovery Call Today!

Do you want to double your income and clients?

Book a Discovery Call Today!

Book A Discovery Call

 

I am Yannis Divramis, I am an SEO Expert. I have been doing SEO since 2013.

I run the Divramis SEO Agency, and I am very glad that you’ve watched this video and keep watching the other videos, because we are posting many videos about SEO every month.

So, if you run a website and want to rank higher in Google, you can ask now for a website promotion offer and get a quote from us.

Ask now for a website promotion offer

See more about SEO:

Leave a Comment