Introduction: A Serendipitous Encounter
Guillaume Pitel: Hello Samir, I’m delighted to have you here today for this interview. I stumbled upon your PageRadar website somewhat by chance, and when we got in touch, I had no idea you were French, I had no idea there was a French SEO professional behind the site.
Here’s what happened: a few weeks (months) ago, we launched the Ibou project, and IbouBot, the crawler for our future search engine, started crawling around the beginning of August. Since we also have another crawler on Babbar called Barkrowler, we regularly monitor our position on the Cloudflare Radar which ranks web crawlers.
And so IbouBot – it had been two months, two and a half months since we’d told Cloudflare: “Hey, we’re here” and they weren’t detecting us.
Guillaume Pitel: So I was trying to verify if we were actually crawling the web properly. I simply searched for our crawler’s name on Google.
Before finding you, I had found a blog post about top crawlers on the WAF360 site, dated August 15th, so just a few days after we started crawling. They had detected us at the very beginning of our crawl, when we had just started, and IbouBot already appeared in position 120. Amusingly, another well-established French search engine was in their ranking at position 125.
Another remarkable thing is that I saw in their WAF360 ranking that Barkrowler was 12th worldwide. That starts to carry some weight when you know that behind you have players like Google Bot, Bing, Cloudflare as well, and ChatGPT, Perplexity, etc. who crawl with considerable resources.
I continued searching for other potential web and crawl observers, and I found your page with the crawler rankings. You had put us in the AI crawlers category, whereas we’re primarily a Search Engine.
I thought “He’s seen us, but the IbouBot page could be improved.” I send you an email: “Hi we’re not an AI crawler…”, I included a link to our manifesto, to additional resources. And you replied right away.
Samir Belabbes: I don’t remember if I replied in French or English, but in any case, I caught that you were French.
Guillaume Pitel: Yes, I suggested doing an interview because I was interested in what you’d built.
Samir Belabbes: I think it’s because the entire site is in English.
Guillaume Pitel: Indeed, the entire site is in English. And so that’s how we met, by chance, because I was looking for why Cloudflare wasn’t putting us on their radar. And I eventually got the answer because after a quick message on Bluesky, that same day we appeared and we were 50th worldwide among crawlers and 8th among search engines. They had forgotten to check the “visible” box…
Well, I’ve talked way too much. I’ve planned in this interview to ask you about different aspects: first, you, your background, how you got into this world, how you’ve also experienced the web in general, search engines. And then, how you experienced the arrival of generative AI in this small world, since you’re young (younger than me anyway :-).
Samir’s Journey: From Personal Development to SEO
Samir Belabbes: Thank you very much for the invitation, the email made me really happy and I always take great pleasure in doing this kind of interview.
To start with my background, I’ve been in SEO for about eight years, and unofficially for a bit longer, so for about 11 years. Eleven years ago, I created my first site that got SEO traffic. It was more of a passion site. After some time, through the number of articles I published, through natural links I had because the content was pretty good quality I think, I ended up getting visits every day: 50 visits, 100 visits, 150, 200 to 300.
It was a site about personal development, meditation, things that spoke to me. And it’s also with this site that I made my first euros. I did a bit of Amazon affiliation, mainly commissions towards book recommendations I would recommend on meditation, personal development, etc. So these were really tiny commissions: 40 cents, 50 cents each time. It wasn’t going to make me a salary, but on the other hand, it was my first euros.
Alongside that, and at the time it was something that worked quite well, I also did an infoproduct: I sold a PDF, which was a sort of personal development compendium that I had enjoyed creating at the time, and which was kind of a synthesis of all the books I had read.
I did this experience for three, four years. I also engaged, I went to Facebook communities to get traffic, to get people to subscribe to my newsletter, to have a kind of snowball effect and have a bit more traffic than just from Google.
You have to know that at that time, I didn’t know what SEO was. I had plenty of visits at my level, but I didn’t know what SEO was. I knew what Google was, but I didn’t specifically know that people typed queries, that there was a certain number of people who typed each query every month within a volume range.
And once I understood that, it was in 2017, just before I started working in SEO. Something clicked in my head. I thought: actually, I managed to get these results somewhat self-taught, I didn’t study for this. So I started trying to learn everything I could about SEO. I remember, at the time, I followed a kind of free module on Moz which was a training by Rand Fishkin over 10 modules. He explained what’s important in SEO: content, popularity, your site structure, internal linking, technical aspects, etc.
And once I understood that, I started applying for job offers in the field. You should know that before that, I was coming out of a year of studies where I had just gotten my bachelor’s degree. I was supposed to do a master’s degree as a Chinese-French interpreter, so I had done a Chinese bachelor’s degree. I was rejected because I didn’t have enough English in my curriculum. At that moment, something happened in my head. I thought: maybe the universe is telling me this career isn’t for me. So I completely switched, I focused on SEO, I looked for a job.
I remember, I applied to several offers, mainly for web editor positions. It was kind of the most accessible position. I sent lots of applications and one of them was at the insurance comparison site MeilleurAssurance which was the insurance branch of MeilleurTaux. They have an office in Lille and at the time, the SEO manager received me, not for this editor position, but more for an SEO project manager internship position.
He kind of took me under his wing. The fact that I already had a site that generated some SEO traffic, he liked that, the fact that I had some practical experience. And I started with the basics: lots of netlinking, making forum links, that kind of thing, making profile links, trying to find opportunities for governmental, educational links, etc. And alongside that: internal linking, going on Screaming Frog, launching a crawl, checking that there were no broken links, 404 links, images without alt tags, etc. Really the basics.
I did that for six months, every day, I had to make between 5 and 10 links per day. Alongside that, I had to monitor competitors: which keywords they rank for, are there keyword opportunities to go after, do they have new content, etc. I did that for six months, I really enjoyed it.
At the end of this internship, I got an offer to stay on a fixed-term contract, so for six more months. And at that time, my former SEO manager, who was more Grey Hat, left. He focused on his other projects. Another SEO manager came who, this time, was White Hat. He came from an agency, really hyper-interested in everything related to site structure, semantic cocoons, content strategy, etc. Doing things properly, having perfect logs, etc.
This allowed me to see, through these two experiences, that SEO isn’t about there being one truth, but that it’s made of nuances. That each project is different, that each site will have its strengths, its weaknesses, and that each time you need to try to focus on the kind of Pareto principle: do the 20% that will bring 80%, and once you’ve done that, focus on the rest.
In total, I was there for a year. Towards the end, after seven months, I was contacted by several people. I had SEOHackers, so I could have maybe gone to work there, I had an offer from them, it was on the agency side. And I also had offers from advertisers as SEO project manager, and one of them was NorthStar Network, which is the company where I worked afterwards.
The Experience at NorthStar Network
Samir Belabbes: There, we get into a different theme: sports betting and online casino, monetized through affiliation. Basically, we provide and give visibility to sports betting operators: Betclic, William Hill, etc. And in exchange, the company is remunerated for each lead it brings, either with a fixed cost, or with revenue sharing each time the user bets online, across about thirty geographical countries: France, but France being a small market for us at the time, it was mainly other markets: Eastern Europe, United States, Latin America, Spain, etc.
This experience started to open my chakras to the fact that actually, in SEO, even being French-speaking like we are, you can potentially attack any country because there are translation tools that exist, today even more with AI, it’s even easier to ensure quality. And that opened my mind to: “OK, actually, if I make an affiliate site project on the side, I’m not just going to do French, I’ll maybe start in French on the main domain and everything, but I’m going to attack /es, /it, etc.”
And that’s what I did at the time. When I arrived at NorthStar, I created an affiliate site called OutilsWebMarketing.com which I let expire very recently and it breaks my heart, I shouldn’t have let it expire. It did comparisons on SaaS tools: email comparison, best CRM comparison, making articles, reviews, promos, etc. I did that for several years.
And at some point, this site started making me more money than at NorthStar. In very short summary, I stayed a year and a half at NorthStar. I left around the time of Covid to focus on my affiliate site. It went pretty well, but during Covid, I only worked from home. I had, like I think 80% of people during Covid, a little depression from staying at home all the time and not having colleagues. So for a year I did that, and I came back a year later to NorthStar as an SEO project manager.
And after two months, the manager offered me to take the SEO manager position. Because I have this very structured side in my head, at the same time I come a bit from Grey Hat and Black Hat, and I also come from self-taught operations. So I have this side where I’m always interested in what’s happening in SEO, I’m always interested in how we position ourselves, in new uses, in AI, in everything that’s happening, in everything that can optimize our work process and our performance.
Little by little, the team grew. At the start, we were three in SEO, and at the end, about six months ago, I had a team of 15 in SEO: SEO team leaders, project managers, a person in charge of netlinking, a technical SEO manager, a profile among the project managers who is more AI / tinkerer, automation.
With everything it’s possible to experience in a start-up that’s launching. I think at the time, when I arrived, we were making maybe 60 or 70k per month in revenue, which is already huge in affiliation. Six months ago, it was 2.5 to 3 million per month in revenue. Everything it’s possible to experience, from beginning to end, I’ve more or less experienced it and it was super formative for me, to launch projects afterwards and be more or less comfortable and at ease.
Learning to Code with AI
Samir Belabbes: Before leaving NorthStar, I left two months ago. The last year, I started training in code. In SEO, even if it remains a passion, after a year the knowledge ceiling is reached pretty quickly, I find. Once you have good bases, good foundations, and that’s what’s important: in technical, in authority and in content. The level, there’s 90% max. After that it’s really details. And after you can specialize: either you specialize in technical SEO, or in SEO authority, or in content. Or you’re a generalist a bit your whole life, you stay a manager. And that’s not a position that interested me and I was starting to get bored. So I started learning to code.
You should know that I had wanted to learn to code for years. Each time I started tutorials, I think the first one was on Site du Zéro, it was on… then on OpenClassrooms, I don’t remember. There were lots of sites. PHP training, SQL training, JS training. Each time, I followed for four, five hours, and I just didn’t understand, and it was so unnatural for me. I told myself: “Actually, I’m never going to make it.” And each time I gave up.
And this time, a year ago, I changed my way of thinking. I told myself: “Actually, I’m going to stay, I’m going to stick with it. I’m going to do two hours every day.” It was Laravel, so that’s the module I followed. “Even if I don’t understand, I’m going to continue every day, I’m going to spend two to three hours until I understand.” I told myself: “It’s like when I was learning Chinese. It’s memorization. At some point, there are patterns that come back, there are application structures that come back all the time, regardless of the language, there are ways of working, there’s syntax that comes back. And by seeing them, I’ll necessarily understand. After a year, if I stay on it for three hours, I’ll understand.”
I followed my little tutorial on YouTube. At the same time, I wrote the same: I followed what the trainer did, I wrote my little code, I had my little local server. I made my little duo: JobBoard. The application and project is a JobBoard that you have to make at the same time.
Since sometimes there were bugs and I didn’t necessarily have the patience either to debug for five hours – although it did happen that I debugged for five hours – I helped myself with ChatGPT. Today, it seems completely obsolete now that there’s Claude Code, etc., directly in your editor. But I copy-pasted directly the error I had, I copied almost my entire file – not the files where there were API keys – but all the files. I said: “Warning, this brick, I have this error, etc.”
And ChatGPT would tell me: “Here, the error, because it could be this, it could be this, it could be this, up to you to see.” So there, I’d go into my file, I’d correct it, refresh on the local server, bam! Ah there, it works. And I did that for two, three months with ChatGPT in a really dirty mode. It’s considered dirty I think today, but I learned to code with it.
Guillaume Pitel: It’s super interesting, we’ll maybe come back to it later. But me, I was talking about it yesterday with my colleague Manu, we were asking the question: how do you learn a new language, we who have great experience with languages, and how do young people, in quotes, manage today to get started in a language? Because actually, the barrier to entry… well it hasn’t really changed, but what has changed is the barrier to achieving something impressive.
We were laughing, he said: “Me, when I started, I made a 3D cube, people behind watching me would say: ‘But it’s magnificent, how did you do it, it’s so smooth!'” And everything today is a snap of the fingers. So making something that impresses, either others or yourself in dev, isn’t so obvious.
And from a certain side, AI can maybe help take those first steps. Because actually, the real key to succeed in starting in a language is to have a problem to solve. If you don’t have something to code, you have no reason to develop something. And one of the first things that happens to several people at our place is: you have a packet of data, you need to process it, how do you do it? You put it in Excel, you do it by hand or you develop something. Anyway, we’ll come back to that later.
Creating PageRadar: An All-in-One SEO Alerting Tool
Guillaume Pitel: You’ve started talking about it, so we need to go into more detail. How long ago did you launch this site?
Samir Belabbes: I launched it in May 2025.
Guillaume Pitel: You have a very broad vision of the field of possibilities around SEO, the web, presence, etc. Can you detail a bit: what does PageRadar do? A list of features. Who is it for? Is it for super experts, is it for the average person? What differentiates it from competitors? And also, the Free Tools you’ve developed on the side for your own visibility, and also to serve the community a bit, especially for people who are starting out, because often people who are starting out need these free tools.
Samir Belabbes: At the very beginning, just quickly on the creation. Just after making my first project with Laravel, which was a personal programmatic project, I was still at NorthStar in January. My boss was a bit tired of always having sites that were unstable in terms of speed. Sometimes we had web pages that suddenly took five seconds to load, sometimes it was an entire section of the site, etc.
And he told me: “I’d really like to have alerts for slow pages, etc.” So there, I thought: “Okay, I’ll try to make something.” He told me that on Monday: “Give me five days.” On Friday, I realized it wasn’t possible to do it in five days. So I thought: “Wait, don’t worry, I worked on it.” For six months, from January to May 2025, I released the SaaS in a state… it’s worse than a beta, it’s worse than an alpha, it wasn’t great.
I iterated little by little. So the first feature was performance monitoring. It’s very simple, I’m not going to say it’s rocket science. I use the PageSpeed Insights API, I use the Crux API for pages or for groups of pages, it can be interesting. So it allows monitoring for any site, plus I think, page typologies. Do you want to monitor your product category typology for an e-commerce? You’re going to monitor five product pages that will be the typical pages, the pages that get the most traffic maybe, your homepage, and all the pages that are interesting for you. So that’s the first feature.
You have your dashboard and day to day, if you have one of the metrics that drops, you’ll receive an alert. With this feature, I understood that the hardest part will be the fineness and granularity of alerts, because you can very easily send alerts that are like crying wolf. You cry wolf, but actually there’s no real correct alert. Core Web Vitals are highly dependent on the user’s device, their country, their network, etc. But anyway, it allows regular monitoring. You connect every day, you see if your site is fast or not, you know exactly which metric is problematic, the style, it’s slow or not. And if you want, you activate alerts for your site.
So there, I started on one feature. We’re in 2025 and Core Web Vitals have existed since 2018. So I thought: it’s not going to be enough to make a tool like that which is just a PageSpeed wrapper. I thought: very simple, I’m going to add monitoring features little by little, and I’m going to try to see what works simply. I’m going to create monitoring features that are useful for me, that are useful for NorthStar, and if there are some that are particularly interesting for users, I’ll push on them, I’ll refine them, I’ll go more into detail and I’ll go further in product quality.
The Different Features of PageRadar
So there, I started adding after that, very simple, like uptime monitoring. I’m going to ping your pages every day or every hour and I’m going to see if they’re in 200, 500, 404, etc. This wasn’t used much for NorthStar because we had a lot of instability on that.
And after that I gradually added page parsing. Every day, it does a diff to see what changed in your pages. That can, for example, help you monitor a competitor, know what they changed on their page and what you can change. Or for example, us, we talked a lot about the operators we worked with: Betclic, Winamax, etc. If they changed something on their bonus offers, sports betting, etc., I get my alert directly. So there, I did HTTP curl, HTML parsing with a diff.
And after that, I added robots.txt checking. It’s a diff every day: what did I add to my robots.txt? I did sitemap monitoring: monitor every day, are there new pages, etc.? Does it respond in 200 or not?
And little by little, I moved a bit away from SEO. I added Reddit monitoring. So each time there’s a brand mention on Reddit or a keyword, a ping and you have a topic that appears. So there, it can help if you want to have visibility on Reddit, which has a contract with Google and which, therefore, is also one of the biggest sites in the world. I think it’s top 10.
After that I did affiliate link monitoring. It’s going to monitor the entire affiliation redirect chain, verify that they’re in 200.
I also did a broken links thing. You can check if you have broken links on a page or on a complete sitemap, and I send you an alert when there’s a new broken link.
The Philosophy: Alerting Above All
And that’s kind of the idea. People who come generally have a feature in mind that they like and they’re going to want to use that one because they really had the problem of: “Oh, my robots.txt, someone changes something and I’m not aware, I don’t know and I realize three days later.” Or “on my website, sometimes I have broken links and I don’t realize it, I realize it two weeks later.” So actually, it’s always people who come in mode: “I’d really like to have alerting.” That’s it, it’s really the frustration and the problem of: “I want alerting and I want to be aware as soon as there’s a problem on one of my pages.”
So me, that’s what I had in mind. It’s starting to get traction. Now, I’m starting to get 5 to 10 signups per day. Obviously, not everyone converts, that’s the hardest part and that’s where I try to refine a bit and contact people and see: why did you sign up? What interested you? Why didn’t you go further? Is the free version enough for you, etc.?
The idea is to make a small SEO monitoring tool that doesn’t want to revolutionize, but that’s focused on alerting. Later, I’ll maybe do backlink monitoring. When a new backlink arrives, you know. When it disappears or someone removes your backlink that you paid dearly for, you get an alert. But always around this idea of alerting, because I think there are many tools that do a bit of everything. You know, Ahrefs, SEMrush, etc., they do ranking, etc. But when a market is saturated, you need to find a sub-niche and be the first in that niche. And me, so, it’s this idea of making an all-in-one alerting tool and each time making it evolve according to what people want.
Free Tools: Creating Value for Users
Samir Belabbes: On the Free Tool part now, it’s, I think, something that’s essential today when you want to have a minimum of visibility and something that allows being easily shared and reshared and that pleases people. Everyone likes to have, when they go to the market, a little free sample, a little sample of sausage or cheese. A free tool, I think, is kind of the same. It allows providing service to people. There are still many searches on keywords: “checker”, “redirect checker”, “status code checker”, etc.
When we share on social networks, generally, people like it, they bookmark it. So, interaction on the page when they come, because often it’s forms, etc., they can configure the thing. They’re going to stay on the page because maybe they’ll reuse it later, they’ll maybe come back because they bookmarked it and they need it in a week. And all that creates user interactions, and that’s not something you can fake. You can do it with third-party tools where users will replicate scenarios: “go to my site, click there, etc.” But that’s artificial and that doesn’t interest me.
I want users from around the world who all have, for each free tool, a different need. There’s one who will extract domains from a list of domains, for example if they’re doing backlinks. And each time, it’s going to be the same use: they come, they scroll, boom, they stay five minutes, sometimes 10 minutes, 15 minutes. And all that is data that at some point will positively impact the site. If you have lots of users who stay 10 seconds, the signal is null. And that’s maybe the case for affiliate sites with basic quality content.
But a site that will offer many free tools, there will really be time on page that’s 5 to 10 minutes with 100 events, 150 events on the page. And all that, at some point, several months, I feel it takes effect and it’s positive in my rankings. It will help me, in addition to quality content, in addition to authority and mentions I make, etc. It’s this somewhat invisible thing that, I think, is super important.
And it requires getting out of the SEO brain a bit that will think more about a checklist, but to say: what can I do that’s useful for people? And the more I do that, the more I feel people come, come back, bookmark, they talk about it around them, etc.
Concrete Example: The SEO News Aggregator
Recently, I made a news aggregator, meaning today SEO news that make blog articles, etc., I took 12 sources and every four hours, I’m going to parse their RSS feed. I’m going to take the latest articles, I do an ORDER BY date, I take the most recent, I take the image, etc. And hop, I display it in my aggregator. And it makes an aggregator that allows SEOs to come and every day do their monitoring on one page instead of opening 15 different pages, etc. It’s a small use, but I arrived in the morning…
The creation process for that, it’s sometimes quite simple. I arrive in the morning, I was too lazy to connect to Feedly that I use, which is also a personalized RSS feed. I tell myself: “I’m going to create mine, I’m going to make an RSS feed for SEOs.” And so there I share, and there you go, every week there are 5,000 people who come on it because it’s really useful for them. And all that goes in the direction of making a tool that’s useful.
So it requires, I think today, when you make a product / SaaS, to really say: “OK, I put my SEO brain aside and I make something useful.”
The AI User Agents Directory
So I made several free tools. One of them is the list and a kind of small directory of AI user agents. I had seen a site that made a list of user agents, I thought: “It’s very cool, but on the other hand it’s less…” You need to make a real directory, I think, with real pages, etc. Because today we have a lot of data on user agents, and we had a bit less data on AI user agents.
So I thought: “I’m going to make a page, I’m going to make a kind of small directory, I’m going to take lots of sources, I’m going to take Cloudflare’s radar, I’m going to take the top 30 that’s on these queries.” And I tried to analyze: OK, what are the different types of bots? What are they used for? Am I going to put a link to their documentation, etc.?
And so, I made my max prompt in Claude. I generated it in Claude, first the desktop version. And after that, I’m in Claude Code, so in my code editor, I have my terminal directly with Claude who I talk to. I gave him the Claude prompt that made kind of the specifications of the directory I’d like to have: different sources, types of bots we’re going to put, filters we’re going to put, etc. And then Claude Code generated it for me, and there you go after several tries of refinement, etc., the page.
So that, each time, I pay hyper attention to the local. I want everything to be perfect, I want the internal link to be perfect, I want it to be graphically good and pleasant enough to read and use. And for IbouBot, on the other hand, I put it manually because the advantage is that I’m alone, so I’m small, I’m not a big company that makes 100 million. So, as soon as there’s a new thing, I can put it in five minutes.
And at the time, it was Sylvain who was communicating a bit on LinkedIn about IbouBot. I thought it was cool. I had just created my kind of small directory of AI user agents that I update every month. I had… I updated my directory with IbouBot. You contacted me a month ago, I think, so I changed the directory with the right description.
And so, each time I try to update and be my little craftsman, let’s say, at a small level. I’m aware that I can’t compete with big SEO tools, but I can make good quality at my level on tools I find useful and on free tools too that I find useful for people. That’s it, that’s my approach.
The Development Philosophy with AI
Samir Belabbes: I think it’s become more and more easy today, as you said before, there’s a barrier to entry for code that’s lower. But the barrier to entry to do quality work, I think it’s still high, because actually, OK, anyone can create anything in one prompt or a few prompts.
Except that today we’re starting to have, I think, in my opinion, kinds of AI patterns. You know when a site is generated with Lovable, etc.? You know, they’re purple sites. There’s always the same color code, there’s always the same responsive problems. You go on mobile and you see that the site is completely shifted to the right. There’s always this kind of thing.
And so, I tried to get out of that and tell myself: “Actually, even if behind the scenes I code with AI, I have my local server, I test absolutely everything.” I’ve been using SaaS for 10 years, and so each time I was frustrated in certain SaaS to tell myself: “Why didn’t they do that? Such feature doesn’t work.” Or “Oh well, Ahrefs! Site Explorer, they’ve had the to-be-launched feature for 10 years.” This kind of thing: “It’s coming, it’s been coming for 10 years, why don’t they develop this thing?”
And me, that frustrated me. And so, now that I’m at the controls of a SaaS, I tell myself: “But it’s too cool, I can do what I want.” I wake up in the morning and if I want to make a new free tool that does that, I’m going to spend three, four hours on it, I’m going to be very careful, I’m going to communicate, it’s going to be too cool and it’s going to make a kind of snowball effect like that, week after week, which is really cool.
Guillaume Pitel: Super nice. Yeah, clearly.
Samir Belabbes: It excites me because I was a bit frustrated staying in SEO and I didn’t see myself staying a project manager or SEO manager for 40 years. So I told myself: “Go ahead, I’ve always loved creating things. Today with AI, you can create what you want” – I think, or at least many things. I told myself: “I’m not going to make the mistake of telling myself: ‘I’m going to make something in a few prompts.’ I’m going to make something good, based on a real framework.”
Laravel, there are many who criticize, they’ll say: “Yes Symfony, why don’t you take Symfony?” It doesn’t matter, I’m going to try to go as far as possible with that. I’m going to try to prove that you can go very far with a simple framework, well, but that allows going fast. Me, that’s what I liked. There’s a nice ecosystem.
They released Laravel Cloud a few months ago, which is a bit expensive, but which has a kind of all-in-one service that manages your server. You have automatic deployments, you have all the CI/CD that’s automatic, you have a command space that you can use directly, it’s super simple. And me, that was the will: I’m alone, I’m going to try to go as simple as possible. I’m going to try, even if I have to pay a bit more expensive for the service, I want to do as simple as possible, as efficient as possible and what allows me to go fast alone.
The Growth Strategy
So there are free tools to have a snowball effect and for people to talk about it. The next step is to have kinds of ambassadors. I’d really like to have local ambassadors in countries, people who will see companies, for example in the United States, in Latin America, etc., to be able, me being almost alone, to grow and do that in a professional and squared way.
Knowing that I don’t have the means of big companies, but on the other hand I’ve always loved the idea of being the outsider. You’re the small guy versus the big ones, and you still manage to take market share. And it’s always what I did in companies. When I was at NorthStar, it was never the top 1, it was always the second or third. And I like that, this idea of fighting with weaker weapons, but still being better. That’s what I like.
The Experience of the Web and the Evolution of Search Engines
Guillaume Pitel: How have you experienced the evolution of the web and search engines since you’ve been interested in them?
Samir Belabbes: So, I really started getting interested around 2007-2008. At that time, Google was already dominant, but there was still Yahoo fighting a bit. MSN Search which was becoming Bing. It was still an era where you could imagine there would be competition.
What struck me most was the professionalization of SEO. At the beginning, it was really the Wild West: you stuffed keywords, you made links en masse, and it worked. Then Google started to become really intelligent with Panda, Penguin, all these algorithms that changed the game.
Guillaume Pitel: And did you see the first AI crawlers arrive?
Samir Belabbes: It’s recent. Actually, for a long time, crawlers were pretty simple: Google Bot, Bing Bot, some specialized crawlers for social networks, news aggregators. And then suddenly, with the explosion of generative AI, we saw many new players arrive.
ChatGPT Bot, Claude Bot, Perplexity, Anthropic… All these crawlers that vacuum up content to train their models or to feed their RAG systems (Retrieval-Augmented Generation). And there, it became much more complex to manage for site owners.
Guillaume Pitel: What’s problematic with these AI crawlers?
Samir Belabbes: Several things. First, they’re often very aggressive. They crawl tons of pages, very quickly, without always respecting robots.txt. It can strain your infrastructure if you’re not prepared.
Then, there’s the question of exchanged value. When Google Bot crawls your site, you know that in exchange you’ll get traffic from Google. It’s an acceptable deal. But when an AI crawler vacuums up your content to train a model that will then answer people’s questions without sending them to your site, there it’s more complicated to justify.
Guillaume Pitel: It’s true that it’s a real current debate.
Samir Belabbes: Yes, and that’s why on PageRadar, I try to document all of that. I give information on how to block these crawlers if you want, how to limit them, but also how to authorize them in a controlled way. Because ultimately, AI is part of the web’s future, whether we like it or not.
The Arrival of Generative AI and Its Impact
Guillaume Pitel: Speaking of which, let’s talk about the arrival of generative AI. How did you experience it as a web professional?
Samir Belabbes: So, at first, I was pretty skeptical. When ChatGPT came out at the end of 2022, I found it technically impressive, but I didn’t really see the impact it was going to have on SEO and on the web in general.
And then very quickly, I realized it was a real game changer. Because it fundamentally changes the way people search for information. Before, you typed keywords into Google, you clicked on links, you browsed pages. Now, you ask a question to ChatGPT or Perplexity, and you get a direct answer.
Guillaume Pitel: And has it changed your way of working?
Samir Belabbes: Completely. Already, I use AI every day now to code, to analyze data, to write content. It’s become an indispensable tool.
But beyond that, I had to rethink my entire SEO strategy. Because it’s no longer just about being well positioned on Google. You also have to think about how your content will be used by AIs, how it will be cited, how it will be presented in generated responses.
Guillaume Pitel: Do you think it will kill traditional search engines?
Samir Belabbes: No, I don’t think so. I think there will be cohabitation. Traditional search engines will evolve, they’re already integrating AI into their results. Look at Google with Bard, Bing with Copilot.
But there will always be a need to go directly to sources for certain things. For purchases, for official information, for specific content. AI won’t replace everything, it will complement.
Using Claude and AI Tools for Development
Guillaume Pitel: You mentioned using AI to code. Can you tell us more?
Samir Belabbes: It’s a tool that has really changed the way I work as a developer. I mainly use Claude, with the code agent. It’s an incredible tool.
Before, I spent my time copy-pasting into ChatGPT, reformulating my prompts, trying to make the context understood. Now, with Claude directly integrated into the IDE, via the command line or via tools like Windsurf or Cursor, it’s fluid.
Guillaume Pitel: What’s the main difference?
Samir Belabbes: Already, Claude has access to all your code. You can do commands like /add to add files to the context, or @web to search for documentation online. And then, it can directly edit your files, create new files, execute commands.
For example, I tell it: “Create me an authentication system with Laravel, with migrations, models, controllers and views.” And it does it. Obviously, I check everything, I don’t let code pass without reading it, but it saves me a crazy amount of time.
Guillaume Pitel: And does it really change your productivity?
Samir Belabbes: Completely. When I switched from copy-pasting with ChatGPT to using the prompt directly in the PhpStorm terminal… I use PhpStorm, I know I should maybe use VSCode but I made the choice to pay for this tool, and it’s remarkable. I’ve gained enormously in productivity compared to doing things manually.
And for developers, it multiplies your efficiency by 10. It allows you very quickly to refactor your application, to quickly find details that would have escaped you. And since you have the expertise, you know exactly what you want to do, how to do it, and you’ll reread all the code generated by Claude.
Guillaume Pitel: Are there other features you use?
Samir Belabbes: Yes, there’s a recent feature called Claude Skills. Actually, you can teach it to do a specific task. It’s like a recipe. You can teach it, for example, via APIs to create a server, to provision it, etc. And it becomes a skill. That way, you don’t have to redo it, it’s a way to automate and not constantly repeat the same actions.
You can even have it do your unit tests. You explain to it what you want to test. I talked with developers, they told me it’s what took them the most time and was the most tedious. Today, by guiding it, it has access to all the migrations, all your models, etc. You have it generate your unit tests, you verify well the quality of the produced code, but if you have the knowledge, you know that what it does will be functional, conform to your standards or not. And if it’s not the case, you nuance with it, you refine and you discuss simply.
Reflection on the Value of Tools and Technology
Guillaume Pitel: There’s just one point I want to come back to before we finish, it’s what you said: you’re not a sucker because you pay for PhpStorm. There’s something very important: technology has to be paid for because there’s time, people worked behind it.
PageRadar, it needs to be paid for, and you need to pay yourself. There need to be people who pay for it. Even if tomorrow JetBrains was bought by Adobe, even if tomorrow Adobe decides it’s free and your tool is integrated into it, what you do is also integrated into it, it would be terrible for you.
And VSCode, the fact that it’s free, it’s a real question to ask: should you absolutely use free tools, or should you pay for tools on which people have made a real effort?
Microsoft can afford to pay developers for VSCode. They want people to develop for what they generate revenue on. They have money on Windows, so they said: “Visual Studio is less and less used in paid version, people don’t pay for the license anymore.” They made Visual Studio Community, they made offers for students, because on the side there were people who offered free editors.
But potentially, when you’re a company, when you make money, you need to remunerate these tools. It’s the same for knowledge, Wikipedia, you need to contribute. You shouldn’t consider that you’re naive because you give to Wikipedia. On the contrary, you enable… It’s the same thing moreover for journalism.
It’s very philosophical, but at some point, you need to ask questions. When the product is free, why? Whether it’s Chrome that collects a phenomenal amount of information.
Monopolies are real problems for innovation. Why? Because precisely with VSCode, Microsoft will maybe put JetBrains in difficulty who edits PhpStorm, who edits IntelliJ, who edits many tools. And them, they only have that, they need to work, they need to live from it.
It’s the same with us, with our tools. There are cheaper tools. Often, when a tool is free, it’s that someone pays behind. It’s investors who put money in capital to try to eliminate the competition, and it’s frankly bad for everyone.
In commerce, you don’t have the right to sell at a loss. You need to ask the question: the fact that VSCode is free, that such and such tool is free… It’s free! Claude has a limited free offer, because otherwise they couldn’t hold financially.
Tool pricing, it’s really an important question. Because if it’s 100 times more expensive…
Samir Belabbes: You’re maybe not going to pay 20,000 euros per month to have Claude.
Guillaume Pitel: If it’s 10 times more expensive, 200, 2,000 maybe. Today ChatGPT, etc., they’re all losing money.
Samir Belabbes: They’re all losing enormous amounts of money.
Guillaume Pitel: And it’s extremely costly in resources.
Conclusion
Guillaume Pitel: Listen, thank you enormously, it was super interesting. I was pleasantly surprised to have made your acquaintance by chance and I’m really not disappointed by this interview. So, bravo and thank you, let’s say goodbye and see you soon.
Samir Belabbes: Thank you Guillaume, it was a pleasure!
