Content automatically, coder-free website, new customer research. Welcome to the AI company: Jan Řezáč (#189)
- Martin Hurych
- před 2 dny
- Minut čtení: 26
Is it possible for a company to generate content, communicate with customers, do research and marketing... without a finger on the keyboard? It is.
And I invited a guy who does this every to the podcast.
Jan Řezáč of House of Řezáč hasn't been deciding whether to use AI for a long time. He's just deciding how. While most companies are still finding the courage to take the first step, he already has his own butler (an AI chatbot), his own talking avatar, and a content recycling machine that churns out dozens of outputs a week - all with minimal human intervention.
I recommend this episode to anyone who's serious about AI - and doesn't want to cut underneath themselves by forgetting the most important thing: who your customer is, what's bothering them, and how to really help them.
We talked about it:
🔸 How does House of Cutter use AI for marketing?
🔸 How to maintain quality and SEO in the quantity of information?
🔸 How do people react to a digital avatar?
🔸 What will the web look like in the age of AI?
🔸 Why it is still important to understand the customer
"AI is not ChatGPT. AI is a bunch of tools, automations and agents that have a bunch of features. And it's useful to have someone in your company who knows about them. Otherwise, you're going to be challenged soon."
Jan Řezáč | CEO @ House of Řezáč, s.r.o
Content automatically, coder-free website, new customer research. Welcome to the AI company
(transcript)
Who is Jan Řezáč
Martin Hurych
I'm Martin Hurych and this is another Ignition. If you want to be kept up to date B2B strategy, trading, innovation and working with people, then consider adding your email to the list of 1,300 existing emails from owners and CEOs of engineering, technology and manufacturing companies. It will help me and you won't miss out on anything I do for you. Today, it's going to be about AI, it's going to be about AI in a very practical way, because my guest is, at least for me and for my linkedin bubble, one of the loudest and most visible faces and advocates for deploying AI virtually everywhere.
Today we will talk with Jan Řezáč, the founder of the family and still CEO of the House of Řezáč. Hi, Jane.
Jan Řezáč
Hi, Martin.
Did he buy a plane yet?
Martin Hurych
I've said several times that I prepare for the podcast using AI, you were no exception because it saves me a ton of too. I used Deep Research on you today, so I'd like to ask if you've got the plane you've been saving up for yet.
Jan Řezáč
That's really Deep Research. I don't have my own plane at the moment, and let's be honest,
that priorities have shifted a bit over the years.
Martin Hurych
I have one more such ice-breaking question, is that you or is that your avatar here in front of me?
Jan Řezáč
Of course it's my avatar. Why would I go to Prague when I can send an avatar? It's little thing that's in the middle of me that projects this 3D thing.
What has changed in Jan in the last 12 months?
Martin Hurych
I think you've scared a few people off, because I can never really tell on LinkedIn when it's you and when it's your avatar, and I'm primarily counting on it already being your avatar, but we'll get to that I'm sure. You were here give or take a year and a bit ago. For those of you who didn't see Jan last time, I highly recommend it because it's one of the most successful episodes. So if you don't know Jan, definitely check out that first episode where he talks about himself at length. Now just tell us what's changed in your company and you personally in the last year and a bit.
Jan Řezáč
Basically, what has probably changed the most is my relationship with large custom design collaborations. I used to think I could do it, but now I think I'm more likely not to do it and much prefer to do smaller, clear, closed collaborations where there's a clear input and a clear output. We've started to move significantly from user experience to AI marketing and AI content work.
How does House of Cutter use AI for marketing?
Martin Hurych
We are talking about AI more and more often here at Zážeh. Many times they are theoretical debates about what could and few of us can show real use cases. You're known for not being afraid of avatar, you were one of the few who stuck with it to this day even after that wave. You're one of the few to publicly say that a lot of your stuff is generated. Let's start by talking a little bit about how you use AI in your company on a daily basis. What I said in the last podcast as well, and what I see a lot, is that a lot of people still can't imagine what real AI can solve, and if I don't know, I can't deploy it. You're a pioneer, a fanatic, what are you doing with at your company at the moment?
Jan Řezáč
My experience is that the moment you say AI or generative AI, a lot of people now think of ChatGPT because it's the most well-known tool within AI. ChatGPT has that major advantage and disadvantage and that is that there's a text box and you can chop absolutely anything into that text box.
Most people don't realize that the box is a very expert box that you need a lot of knowledge to pour into it to get something that then gives you that useful result, and then they're dissatisfied with what they're getting. That situation is getting better because even OpenAI is recognizing that situation and therefore adapting to the fact that people are writing one-sentence commands there and enriching those things internally to give them some reasonable output. Anyway, if I had to take the first thing when you say using AI in a company, get rid of ChatGPT. There are use cases, there are scenarios, there are places where you can effectively use AI tools already and most of them are not ChatGPT.
Martin Hurych
You broke a lot of people's toys, because when you talk to people, AI is ChatGPT. For a very small fragment it's some Claude and other LLM models, for experts within the population 's some pictures, and now you're saying to forget about it and still look somewhere else.
Jan Řezáč
If I were to expand even a little bit on the possibilities that exist today, AI models are still evolving. We're filming in early March and what we're seeing in that market is that that evolution is slowing down, that those models are getting bigger and bigger and at the same time they're getting more and more expensive. The jumps that were in the last 2 years are not as extreme as they were. It's slowing down perceptually, and at the same time there's a huge amount of different specialized tools built on top of those models now, we mentioned that AI avatar for example.
The AI avatar for us at the moment is created by taking some text and converting that into audio. You convert it to sound in a tool called ElevenLabs, because that's the best one at the moment out of that wide portfolio of hundreds of sound and soundtrack tools. Suddenly you have a soundtrack that speaks in my voice. We then throw that into HeyGen, which is a tool that creates that avatar for you based on a five-minute video shot on your iPhone. I'm kind of red in that last set of avatars, I'm going to have to powder myself and shoot some more, but at the same time it's not a problem at all.
So we're able to take those individual tools built on those big language models and plug them into automations. All of a sudden, for example, from an AI marketing perspective, that allows for very fast recycling and new content creation. For example, I can take a newsletter that I've written, have the AI rewrite it to look like spoken word, throw that text into ElevenLabs, it throws it into HeyGen, and at the end I have a video that I can use on Instagram, on YouTube Shorts, on TikTok. I didn't have to intervene at all, yet it all came together. Of course, part of that automation could theoretically be that it takes that text and then it asks some questions, does some research on it, does some research on what could be talked about. It'll come up with 5 more texts and from those it'll make another 40 videos and all of a sudden we can publish 4 videos on TikTok a day without me seeing a single one of those videos.
That's what's happening on the internet right now. Europol estimates that in 2026, which is 9 months from now, over 80%, 90% of the new content that will be generated on the Internet will be generated. Few will be writing, few will be making videos of that total percentage of that content. They're addressing it from a misinformation perspective because that's absolutely critical, you can find anything on the internet now and it's going to be more challenging. We're addressing it from the perspective of some visibility, from the perspective that you as a brand want to be
in touch with those potential customers of yours, you want to be , you want to be credible to them. You want to have some kind of relationship, but you don't build that by them seeing you once and then 3 years later they remember. We're in B2B, we've got long buying cycles, long delays between when something gets decided and when those companies kick in, before it happens they're addressing what you can help them with. Nobody remembers that, they don't have a chance, even if you show up ten times, they still don't remember. You have to remind them and at the same time by reminding them it has to be interesting for them, it can't be interesting just for you. That's been my big life lesson over the last 25 years, that I have to look for those things that are interesting to me because otherwise I'm not enjoying it, but at the same time they're interesting to somebody else.
Aren't we turning off the internet under the pressure of (mis)information?
Martin Hurych
This leads me to the devil's question, because a few years ago many of us turned off our televisions, didn't even get one. Isn't that how we shut down the internet?
Jan Řezáč
Now consider what will happen, an internet overwhelmed by a huge amount of generated content, which will also be tailored be found and wanted by Google as much as possible. That's logical, because if Google can't see it, nobody can see it at the moment. The AI search engines will look at it, of course, but how will the AI search engine know that you're a relevant source?
Martin Hurych
Today, according to Google, according to the catalogues that are created somewhere, the question is whether these catalogues will start to build themselves.
Jan Řezáč
I suppose so. You see it the moment you talk to those AI search engines like Perplexity. You need long articles on a specific topic, long videos on a specific topic, which leads to the fact that you as an organization suddenly don't need to have a 10-page website. You need to have a 10,000 page website and we're back to generating content. We're in a perpetual cycle of you need a bunch of content to be seen and therefore you're creating an even bigger pile of content than the other guys and the other guys are also creating a bunch of content to be seen and it's cheap to create a bunch of content.
How to maintain quality and SEO in information quantity?
Martin Hurych
This puts you in some kind of hamster wheel from which there is no escape. You can see that today, if you just do some still old-fashioned manual googling research, a lot of those articles don't bring you much. In this deluge, compounded by AI, how do you get out, have that massage of quantity, but at the same time not drown in it qualitatively? We've both seen and still see millions of courses that tell you, in 10 minutes you've got a publishing plan, in another 10 minutes you've got a LinkedIn post for next month. But when you see those prompts, the prompt is, write a LinkedIn post about something, but you can't stand behind it and it doesn't give anybody anything.
Jan Řezáč
That's exactly it. The moment people take AI as ChatGPT, the knowledge of what's going on more depth isn't there, so I logically slide to simply talking to ChatGPT. Just adding some prompt engineering to it makes it better. When you give it enough context just in those instructions in that GPT, it'better. You can play with it at that level and you can play with it at that technical level because those big language models don't solve facts, they don't solve statistics, but you can help them solve facts. You plug some of your database into it and it could start to get technical and we're not going to get into the technical stuff so that the outputs are more accurate, but they're also useful those people. Because ultimately people don't if they're reading human generated content or human written content, as long as they don't recognize it and as long as it gives them something.
My answer to this, I'm doing my research via Google, is in the topics I'm interested in at the moment. Start following the people who are good at it, and that's hard to spot, it's a gradual process. You can get a lot of nice, interesting and useful content through Bluesky, which is currently a replacement for Twitter, or even through that LinkedIn thing, or through newsletters that different people write. But you're going after specific people, you're going after specific authors or specific brands that are writing something. You don't go to Google, you don't go looking for 12 simple tips on a topic, that doesn't work at all.
There are these closed gardens where someone tends to some content and makes sure that it is of good quality. It doesn't really matter if that content is created by linking to different places on the internet or downloading relevant articles and videos and putting them all in one place, or if they write it themselves or generate it through AI. There's that guarantee that if it's there, it's good.
How do people react to a digital avatar?
Martin Hurych
You said that people don't care if it's AI generated or human generated, for me it has to bring something to them and even I think the younger ones won't care anymore, even if they might be able to tell it's generated. As long as the information they want is there, it won't matter.
I'll second that, you're turning from blog author to more of a curator of information, and you'your name behind the fact that what's on your site is some opinion of yours, and it doesn't matter where it comes from.
I would go back to the original one, how you use it and what the implications are. We said here that you're still using an AI avatar, there was a big wave of HeyGen that I didn't catch, and now I don't have one because I see that wave of HeyGen has de facto sailed. How do people react to it, for example, when they know or don't know it's an avatar but it's still basically the same character looking the same? What kind of feedback have you gotten from the market?
Jan Řezáč
At this point, a minority of people will recognize that it's an avatar, let alone know that avatars exist. Of that minority, typically people ask how I did it and if I use HeyGen. It's simply still just the video that pushes it forward. We see that on TikTok as well, if you look at TikTok House of Cutter, we have a bunch of videos that are either done by avatar or they're cut videos again automatically by AI. It's never been cut by a human, it's all been cut by AI, in this case it's called OpusClip and there are cut videos and it doesn't matter if the video is an avatar or a human, it's the content that's important.
Who is the butler in the House of Rezac?
Martin Hurych
They work better for me than the ones I cut myself. We use the Klap App, but it doesn't matter. I've invested dozens of minutes many times to find the thing, and then you throw it into some tool, it calculates virality, and those videos have better viewership than what I'm picking up here long term. So we discussed one thing, at your company you generate voice and video from the written word. Is there anything else that you're proud of or that you would rat on yourself for doing through AI?
Jan Řezáč
I'm still excited about our AI chatbot, which we call the butler because when we're House of Cutter, we have a butler. That's again built on an AI tool and we've supplemented it with quite a lot of our information. We've tweaked it so that maybe it talks about some things and doesn't talk about some things. We don't want him to be out there telling people that we're going to give them a discount, we're going to do something for free, that sort of thing. It's clearly written in the terms and conditions that the butler can't offer you anything. Anyway, people interact nicely with him and by them writing in that chat, we can see what's bothering them. We're able to start adapting how we communicate and where we communicate to what's really bothering those people. Because ultimately it's not about what I communicate, but it has to bother somebody, it has to be addressed. So we have permanent feedback from the market and even with our website, where the traffic is relatively low, those people are interacting with the chatbot. It's definitely going to be skewed by the fact that I'm writing about chatbots, so then they're going to try it out and we're even seeing it in those chats that it was probably a fan. But there's also regular questions and needs and we see different types of haters and people trying to hack it, but that's just the way it is, it's a new touchpoint, it's fun, so let them play with it and it's fine.
Martin Hurych
What did you feed it?
Jan Řezáč
First of all, we fed it into our website because one of the things you can work with very quickly, very easily, very efficiently, is the information you already have. Downloading the whole site and throwing it into an AI tool in some reasonable format is no problem today. We've also fed it with other internal information about the products, about how the collaboration with those clients, for example, is going. Because a lot of them are interested in how the collaboration is going to work, what's going to happen there. Gradually, as the questions get more and more, we're not really worried that the chatbot won't answer everything, because that's . What we're concerned about is that in the future those people find it and therefore don't go to the chatbot at all, or that the moment someone asks the question, the chatbot answers it. So we're gradually tweaking and improving the whole thing.
Martin Hurych
I've noticed that you're excited about it because you've written about it a few times on LinkedIn. So that's a feat at the moment, when I come to your site I'll see an intelligent chatbot.
Jan Řezáč
You'll see an intelligent chatbot. For example, my friend Radka said that the chatbot never brought her anything interesting, it never solved her problem. I actually understand that because at the moment there are a lot of even diverse chatbot solutions out there, some are built on humans, some are built on AI of varying quality and it's hard to implement it in a way that's beneficial. On the other hand, I see it similar to navigation. When you come to the web, ideally you don't want to use navigation, navigation is a clue at the moment when I'm not getting what I want. The chatbot is a second such clue so you have a chance to get what you want. For us, it responds so that people actually get that chance and get the information that they came for.
What will the web look like in the age of AI?
Martin Hurych
When you talked about feeding the chatbot with websites, I thought of one more thing. I recently watched the AI Summit in the UK and the opinion is starting to grow here in the Czech Republic that the look of websites is going to change dramatically. You used to make a living by making nice websites optimised for people to find where they need what. Now they say we're going to texting to make it as easy as possible for AI to read. How are you going to do business?
Jan Řezáč
I've already mentioned that we're gradually moving away from websites to other forms of smaller AI products. But I don't think websites are going to die. You mentioned it yourself, a company needs a website, a company needs to have those texting sites there somewhere, so that it's then visible in that AI search, so that it's visible in that deep research, so that its branding is heard.
Martin Hurych
I'll take it practically. What should a website that survives 2026 look like?
Jan Řezáč
The first thing I would on is still customer research. Get to know your users better, which after all can be done through that AI chatbot for example, because I can see what people are typing in there and then I can tailor the future design accordingly. The second aspect is that I need to have a technical solution that will evolve over time and when an AI breakthrough comes, the technical solution will handle it. In other words, forget about WordPress, forget about installing some tiny system from some tiny agency. Because what I can 100% guarantee you is that they can't handle the speed of the change that's happening at the moment.
Martin Hurych
You had some pretty fundamental information here the other day about the fact that you don't have a website worth millions, but that you have it hosted on a platform somewhere. Can you own up to what platform that is?
Jan Řezáč
It's Webflow, but of course there are more platforms, anyway the platform has to keep up
with the times. The third aspect is that at the moment, creating a new website is actually relatively quick. If I as a company am prepared and I'm clear on what my services are, what my products are, what resonates most with those customers, and I have all that information digitally, it's easy. There are AI tools for all of this, it's a matter of 10 minutes to produce wireframes, what's more difficult is taking that know-how of that company and populating those wireframes with that. For smaller and medium sized sites, Relume is a good tool.
Martin Hurych
Have we forgotten something?
Jan Řezáč
I have the know-how, I know my customers, I have a technology platform that changes with the times. Then, based on those things, I have some set of automation that allows me to work quickly and efficiently with that website and all that marketing communication. The moment I have that information digitally and I have it in a good structure, I'm able to generate new landing pages easily. I'm able to easily create basically a set of, here's where I have a campaign, here's where I have ads, here's where I have a page that it leads to, here's where I have a bunch of supporting content that I direct those people to. The whole thing works, and at the same time, most of it comes automatically.
Why is it still important to understand the customer?
Martin Hurych
You've said a terribly important thing here, you need your know-how and your knowledge of your customer and that maglajz in the middle, that's where the technology and the speed at which you create it just changes. So you're not being fatalistic and saying your role is going to end because you're the one who needs to understand your audience. AI will help you in that partly, but you still need to understand it and you still need to be the one tuning what to generate for those people for that content. After that, it's just a change of tool, whether you're typing it once in a , once in T602, or using AI. Do I understand that right?
Jan Řezáč
You got it right. So the typical dialogue with companies is that we're not going to do any research, that doesn't make sense, we'd rather use these three ideas here that we came up with because they're great. But manager number two will say they've been trying this for the last 5 years and nothing has come of it yet. People need to realize that if they want to influence other people's behavior, they have to start with them. At the same time, the research opportunities that are happening right now because of AI are giving me very interesting opportunities to understand these people. I don't have to do as many interviews, I don't have to do as many surveys. So I have new possibilities in getting to know these people.
At the same time, another type of research activity that I can do now is carpet raids. Because suddenly it's really easy to create things in electronic space. It's really easy to take a palette of Rezac's newsletters here and make 60 videos out of them and see if anyone responds to it or not. If they don't, we've lost a couple of hours of work that Rezac wasn't involved in at all. If it works, it works, and we're going to continue to do that and figure out ways get that type of content out to those people more. We're going to see people jumping out of the climate change box like devils, so we're going to do more climate change stuff if we want that engagement, which is not necessarily what brings in the real business. I suddenly have options that I didn't have before because to do all this stuff I just said, I used to need a set of professionals who were slow and expensive. I can suddenly create things here with a set of automation that I couldn't before.
Where does AI (not) help with marketing research?
Martin Hurych
I still see the red line there that you have AI as a tool. A lot of people will tell you why would I bother with some research today when I ask Chat what my client actually wants. That's why I wanted to ask when you're not going to have anything to eat?
Jan Řezáč
People get annoyed talking to people so they don't do it, it's like making a phone call, it's hell. That Chat looks like it could be it and we've obviously tried generated research and we've tried semi-generated research. We've done some set of interviews and we've doggerelled the rest. The basic problem with generated research is that you can't trust it. You can't trust him at all, you have no idea which bits that he's generated are reality and which are not, and it just messes with your head. You can't trust him, which is a problem, because why else would you do research.
What may work already with those current tools is that I'll use ChatGPT or its deep research feature or Perplexity or Claude to expand my idea of what all those people might want to address. It expands my idea before I go do the research. Of course, all of these tools can help me analyze the research data without any problems, and again, there it depends on how good the instructions I put into it, how good the input data is, and what requirements I have of the output. If I play around with this, I'm able to use Claude or use ChatGPT to analyze maybe 20 interviews and I can do it like this in 10 minutes. But there's another catch. moment I do it this way, then I come to the client's managers, supervisors, whoever, and they may not believe it again. Because there's a big difference when you're out there touching the people, when somebody does the interviews, writes out the quotes from them, and then you analyze the data, maybe in a workshop or you participate in the research yourself. You then look at those people and you don't believe your ears what is happening or not happening, how they think about it, because we all have a box that we think in. This expands that box and it expands it for those managers. You need people to start thinking differently in the organization and then from that, how that organization changes its behavior can influence people.
Martin Hurych
That said, we still fortunately need to understand who we are delivering a service to more some ones and zeros. We still don't have enough information about ourselves to understand that, and the statistics are relentless and can take you somewhere a little bit different at any time.
Jan Řezáč
At this stage of generative AI development, it certainly doesn't help you that much to talk to ChatGPT about your customers if you haven't fed it data about your customers first. You have to get that data somewhere, and that data has to be collected in a way that makes sense. The moment you do that and the moment you have that database of 50 interviews with customers that you just uploaded because transcription is not a problem, analysis of that data is not a problem, then suddenly you know more.
How does it explore tools for deployment in practice?
Martin Hurych
I guess it's fair to conclude by saying that a customer in New York is not a customer in Trebic, because the AI fanatics will tell you that there's already enough data out there that you can on it. I'm going to turn the page a little bit and come back to your company. Keeping your finger on the pulse of the times with AI is extremely challenging. How do you organize that in your firm, where you know you're just exploring at the moment, and where you're not afraid to put that newly explored tool into all the features you're building for yourself and for your clients?
Jan Řezáč
There are several aspects to this. One is that you have to make time to do the research at all. We have it created in several different levels and we have it created continuously all the time and we are exploring. By having web analysts, we have designers, so I see more into those designers at this point and how we work in that design team. That's where we've got it done so that each person is researching and we have research directions, so somebody's researching more chatbot, somebody's researching more automation, somebody's researching more vector databases.
At the same time, as we explore, we find out new and new things, so we share and talk about them and enrich each other. I found a beautiful video today that explains in very human language the difference between when you just talk to a big language model, when you enrich it with your data through the RAG method, and when you use fine-tuning. You fine-tune it so that it works better for some cases, so that it's really your customer support and it's not just a generic model and it responds the way you want it to for some situations. I've shared that video, some people have already seen it, and here we go.
Martin Hurych
How do you do it practically? If somebody was looking at us right now, and we've got them excited and they wanted to implement something like this into the company on a practical basis, do you have it through regular meetings, do you have it channeled somewhere in Teams or Slack or what have you?
Jan Řezáč
One thing is the channel set, we don't use Slack, we use Signal. The second thing is that we have an hour-long meeting every 14 days or so, where we enrich ourselves on a topic. My colleague is now going to do a meeting on automation and it's going to be more advanced stuff than she had in the meeting before, so we're going more in depth.
The third thing we do, which we've been doing for a long time, and which is terribly expensive and absolutely hellish, and at the same time keeps the ability to change in that organization, is +1. +1 is that every 2 months we dedicate that last week to internal stuff. One day we reflect on what happened on projects, what learned, how to learn from our own mistakes or what went well. We dedicate one day to planning and the other 3 days and we dedicate those to internal activities depending on what's needed right now. +1 was two weeks ago and we dedicated it all to internal projects and developing those new AI products. The benefit of that is that creates that culture where it's normal to introduce new things, where it's normal to change, where it's normal to absorb something that wasn't here 2 months ago and you're still doing it. The downside is that it's damn expensive, and as a business owner I'm concerned about both.
Martin Hurych
It's awesome that once every 2 months you completely get out of the business for a week and go negotiate. Who's ever gonna say what's on? You research, something comes up, you talk, there must be a prototyping phase of some sort, and I assume you're not the only one with your thumb on everything anymore. Who's going to say, we're replacing tool A with tool B and completely overhauling this automation because we believe tool B is better than tool A?
Jan Řezáč
It depends. Most often it is done by my colleague Marek, who is our lead designer and with whom we have been working for a very long time. I do it, it's done by people who are in that level of technical detail. For example, when we pick an automation platform that we want to run on for a while, like we picked Webflow a long time ago, those people are then able to tell us what direction we need to go. When we talk about new products, we try everything out on ourselves. The moment we tweak an AI chatbot for ourselves and looks like it solves some problems for some part of the market and its interesting to do, then we let it out.
Does he have a CAIO at the company?
Martin Hurych
Do you have a Chief AI Officer type position? We're not talking about polished positions at all right now, but rather if you feel it's better to have it distributed across the team, at least across a company of some size, or if there should be an AI ambassador position, like Filip Dřímalka.
Jan Řezáč
The Chief AI Officer is a position that has power, and that's good to me if I want to move some structure that's kind of resisting that, which is most people. If you're talking to somebody about AI, there are concerns, and those concerns are perfectly logical. For a lot of white-collar professions, there's going to be what's definitely going to happen in the next few years is that they're going to take a script that you used to do and automate it. You were used to it being done somehow, and yet you have to click these things here in this browser and it can't be done automatically by somebody. It can and it will and everybody will be looking at some internal cost savings and the dumber companies will be looking at it in the style of, let's save money on people. The smarter companies are going to create more things, they're going to create them faster and they're going to focus on the work that AI can't and won't do at the moment. In the future, maybe it will be relatively little work, and maybe we'll see companies that are pure AI. This may logically imply in the future that we will each have our own company because it will be impossible not to have one.
Martin Hurych
That's a big question, I've also heard that somewhere you get unconditional comp instead of unconditional income. But I don't think it's going to work completely according to the Gaussian curve and how we are born and how intelligence is distributed. They'd have to change us first.
Jan Řezáč
As for the generational breakthrough, it's going to be even cooler. We're gradually getting to a world where it's normal not to have children, globally we're getting to a world where it's normal not to have children. There was a cool article in the New Yorker about this just now. We're slowing down and we're getting below 2.1. South Korea is such a nice model example of what happens when you have a city that celebrates the fact that the first child has been born in 5 years and the parents are seen as heroes.
Martin Hurych
If I remember correctly, once it drops below something like 1.6, no civilization has been able to recover to its original numbers.
Jan Řezáč
They have 0.7 and it's going to happen, so AI may be the salvation of our social system and other things. At the same time, that transition from how we operate today to how we're going to operate in 10, 20 years is going to be pretty cool, and that ability to change and invent these things in a smart way is going to be more and more interesting.
And when will he have his own robot?
Martin Hurych
As an enthusiast, when are you going to have your own robot? When will your butler materialize into a physical body?
Jan Řezáč
I'm not thinking about anything like that. Every once in a while, something pops out at me from my social bubble that I follow on Bluesky, like here's a dude bringing in a grocery and two robots sorting it where it's relevant and looking humanoid, but at the same time it doesn't make any sense. Why should we adapt everything to be humanoid, including our robots? In some cases it makes sense, in a lot of cases it doesn't make sense, it's going to change and it's going to be slower, changing software is a lot easier than changing hardware.
Summary
Martin Hurych
But nothing software can put yogurt in your fridge. Maybe we'll have a session on that in another year and a bit and see who's already got one of those robots. If for some reason people were skipping and they missed something and now they stumbled upon this site that's supposed to make them, like, listen to the podcast again or something that completely explodes your engagements and likes and comments on TikTok, what would that be?
Jan Řezáč
Ladies and gentlemen, please note that AI is not ChatGPT. AI is a bunch of different models, tools, automations, agents emerging at the moment, assistants that have a bunch of specialized functions. It's very handy to have somebody in your organization to navigate what's going on, because at some point in time, you're going to be challenged if nobody is navigating.
Martin Hurych
Jane, thank you so much, that was great. I wish you to stay on the cutting edge of AI and your field, and I dearly wish I could discuss such news with you on a regular basis. Good luck.
Jan Řezáč
Thank you.
Martin Hurych
If you were a little skeptical in John's final word, it was certainly not a task, on the contrary, take it positively, we have tried to show you the status quo and the immediate future here today. As we correctly noted in the piece, we're recording in early March, so when you see that, you'll see how much we've gotten lucky or whatever has come out in the meantime, because you really don't know the day or the hour at this point. If there's anything stuck in you, John and I did our job well. I have a few requests and recommendations at this point. I haven't told Jan here, but I'm going to knock some bonus out of him so that you have that kind of guidance material on how to approach AI within the company. I'm sure it won't be about tools because that bonus would be 14 days old, so it's more on principle. At the moment, that bonus is already hanging around. If you want to be supplied with information on B2B strategy, business, innovation and working with people, consider adding your email to the list of 1,300 emails from owners and CEOs of engineering, technology and manufacturing companies too. If I've completely charmed you and you'd like to help me through the Internet, consider giving a like or subscription to this episode or if you happen to forward it to someone who might benefit from it. I have no choice but to keep my fingers crossed for you and wish you success, and not just in implementing AI, wherever you can today, thank you.