Trending Articles
Trending Topics
Trending
Matej Kukučka knows what it means to create effective marketing. With a lengthy career spent turning complex ideas into campaigns that can truly connect, he’s led marketing efforts across industries, from established brands to smaller startups. Now working as CMO at Luigi’s Box, he’s focused on combining creativity with data-driven precision, and leveraging AI to give his team the insights they need to stay ahead.
Matej’s approach to AI is rooted in its practicality. Rather than chasing the hype, he’s looking for ways that AI technology can deliver tangible value. “For the recurring tasks,” he explains, “I have started using AI for various purposes”. Whether it’s generating first drafts or speeding up research, his focus remains on saving time without risking quality.
For Matej, it’s important that the efficiency AI can provide doesn’t come at the cost of human insight. He believes that AI tools can create a powerful starting point, but the best results can only be achieved when a human is involved.
This measured mindset runs through his philosophy. Matej knows the value of AI tools, but thanks to his years in the field he knows something more – when not to use them. As he puts it, “AI is a great enabler, but without context, strategy, and human judgment, it often leads to mediocre results or even harm to the brand”.
It’s Matej’s balance of pragmatism and delegation that makes him ideally placed to answer our first question, just how effective has his balanced approach been to his work over the past few years?
The AI has significantly shaped my approach to digital marketing over the last year, but it wasn’t until 2025 that I began to see its substantial impact. Until now, I was certain that I would be doing a similar job for the rest of my life; however, I am no longer so sure.
My approach to marketing is a combination of recurring tasks and one-time events. For the recurring tasks, I have started using AI for various purposes, such as translations, creating content briefs, verifying content adherence to brand guidelines, programmatic SEO and essentially anything that occurs regularly.
For one-off tasks like coding custom lead magnets, creating high-converting landing pages, or sales decks where the output needs to be perfect, I am using AI cautiously. The famous quote about having 6 hours to cut the tree, where you would spend the first 4 sharpening the axe, resonates with me greatly from an AI perspective. The time spent searching for the ideal prompt that yields the desired result is often so high that I would do the work myself more efficiently.
From the CMO standpoint, I still need to be the one who sets the direction and priorities. I feel like the demand from specialists is higher, and not everyone is really ready for the transition. With AI, the demand for higher output is apparent; however, it’s evident that AI cannot replace the human touch. Therefore, I am confident that AI will help those people who can utilise it in their daily work and replace people who won’t adapt.
The successful AI tools are those that make your daily job more efficient. You can do it faster, or the output is better. I believe that AI can enhance speed, but the quality aspect is questionable.
Frankly speaking, most AI tools are simply AI wrappers with built-in prompts. I don’t blame them. However, the hype around them is often not justified, in my opinion. For most one-off tasks, I use ChatGPT, Perplexity, or Claude. Our development team primarily uses Cursor and Luma in most cases. For the trending “vibe coding” of simpler applications such as calculators, we have tested Lovable, Bolt, Replit, Spring, Base44, or MiniMax Agent. Lovable is my preferred solution.
For content AI tools, I like Frase for content briefs and AI generation, Ahrefs’ latest AI features for keyword translations, or their Content Helper to target the right intent. Additionally, I utilise Cuppa for programmatic SEO creation, SurferSEO for topical maps, ElevenLabs for bulk voiceover creation and Gamma for building presentations more efficiently. There are not many tools that I have adopted thanks to the latest trends, but rather, most of the software I was already using has built AI directly into it.
One example I would like to highlight is keyword clustering. When I started around 15 years ago, I was exporting keywords from various sources, including Google Keyword Planner, Google Webmaster Tools (now Google Search Console) and various SEO tools and keyword suggesters. Following that was the lengthy process of semi-automated clustering with OpenRefine. Nowadays, it’s a lot easier. However, ChatGPT or other LLMs have a problem working with larger amounts of data. Additionally, both approaches group keywords based on queries. I recommend using SERP-based clustering software dedicated to this use case, and at this point, my recommendation is Keyword Insights.
This example demonstrates that marketers must know which tool is best suited for the specific problem they are trying to solve. AI tools are very ambitious. They will never tell you they don’t know the answer, even though they are not always right. And that’s where I see the future of digital marketing. If you know what you’re doing, you’re a few steps ahead of the rest.
I believe that AI doesn’t make traditional methods obsolete. I feel like it’s complementary to them, not a replacement. We need to understand why people behave in a certain way. The AI is as good at predicting customer behaviour as the quality of the data. Low-quality data will result in low-quality predictions. Simple as that.
As an example, I can mention the latest HubSpot connector for ChatGPT. It’s less than a month old, so we haven’t had much time to thoroughly explore it. However, early results suggest it’s useful for high-level strategic questions, though its capabilities seem limited beyond that. There are numerous additional activities associated with predicting customer behaviour, including A/B testing, customer segmentation, surveys, user testing and many more. For all of them, AI will be most helpful when you have high-quality and unified data.
When it comes to product discovery, the core offering of Luigi’s Box, the story is different. Our AI models are trained on various signals, including clicks, searches, session patterns, past purchases, product metadata and real-time behavioural data. This depth and diversity of inputs allow us to deliver highly relevant, context-aware recommendations and search results, making AI not just efficient but truly impactful in guiding customer journeys and boosting conversions.
With all these combined, along with collected user preferences and data sharing from other tools (such as CDP and email marketing software), the difference is noticeable. However, we thought that at this point, utilisation of semantic search would be higher, but that’s not the case. People use traditional methods for searching for the product (e.g., search bar on the website) to look for concrete product types, and AI is used for the initial research.
This question has two coins. One is the product I mentioned in the previous question, and the other is how AI helps attract visitors and convert them into leads and customers. The client-facing solutions driven by AI are usually served to a much bigger audience compared to marketing platforms that target prospects of these solutions.
With Luigi’s Box, where we store billions of data points, it’s easier to deliver more personalised experiences. With B2B/SaaS websites that I typically work with and campaigns that target smaller audiences, it’s challenging to utilise AI to achieve meaningful outcomes.
With that in mind, it’s still possible to utilise AI for personalisation. First of all, you need to understand your audience and ICP. A good CRM quality is a prerequisite for letting AI do the work.
One example I’d like to highlight is outbound sales. We collect a broad mix of data, including first-party signals from our website, CRM insights and email interactions, as well as third-party firmographic and intent data. This feeds into our Target Account List (TAL), which is a curated selection of companies from our broader Total Addressable Market (TAM) that we believe are the best fit based on our Ideal Customer Profile (ICP).
We use tools like Clay to automate data enrichment and trigger personalised outreach at scale. This setup allows us to segment effectively: AI handles customised outreach to smaller, qualified prospects automatically, while our BDRs focus on high-potential accounts within the TAL. It’s a smart balance — automating personalisation where possible, and investing human effort where it matters most.
The number one factor is relevancy. If you can stay relevant and target the right accounts at the right time with the right message, it doesn’t matter whether the communication was created by AI or a human. Authenticity in communication often begins with understanding — and that’s where AI excels. By analysing behaviour, preferences, past interactions and intent signals, AI enables us to deliver messages that feel timely and genuinely helpful, not random or intrusive.
AI is driven by data and facts. It helps eliminate bias and subjectivity, and it reduces human error, because while humans have their limits, AI is constrained only by the quality of the data and the power of its processing. However, true personalisation extends beyond simply inserting someone’s first name into an email. AI enables much deeper contextualisation, such as referencing a product they’ve viewed, suggesting content based on their recent activity, or even anticipating their next action.
Most of the AI tools I use allow you to adjust the tone of your communication, whether you want it to be friendly, professional, casual, or persuasive. This flexibility enables the matching of messaging style to the audience or channel, adding a layer of emotional intelligence that enhances authenticity. When you combine this with relevant content and perfect timing, the result is communication that feels thoughtful and personal, even when delivered at scale.
One thing I’d like to highlight is consistency. We utilise our custom GPT app to review all outgoing content, ensuring it aligns with our brand guidelines. This helps us maintain a consistent tone and style across all channels, which is crucial for building trust and authenticity over time. Consistency combined with relevance forms the foundation of effective communication.
It’s also important to remember that while marketers, designers, or developers may obsess over technical details, such as using the latest framework or achieving pixel-perfect layouts, what truly matters to users is whether they can quickly find what they’re looking for and feel understood. This is what I often tell people I am managing. AI helps bridge that gap by focusing on what the user actually needs at the moment. That’s where authenticity really comes to life, not in perfection, but in making people feel like your message was created just for them.
If you had asked me the same question last year, I would have said that it’s easy to distinguish between AI-generated content and the human experience. I am not so sure anymore. To me, authentic marketing is the ability to stay relevant while producing high-quality content simultaneously.
Let’s divide this question into different formats. There is text, audio, video and images. There are obviously other formats AI can produce, such as landing pages, emails, chatbots, workflows and interactive content like quizzes or calculators. However, these are essentially combinations or applications of the core building blocks: text, audio, video, or images. By understanding how AI impacts these fundamental content types, we can better assess its role in delivering authentic experiences across all channels.
In text, AI has come a long way. From my own experience using tools like Frase and Cuppa, I’ve seen a significant improvement in the quality of output over the past year. What used to sound templated or flat now often reads as well-structured, engaging and even brand-aligned with the right setup. Authenticity in text doesn’t mean every word has to be written by a human. It means the content is meaningful, accurate and crafted with the reader’s intent in mind. If it delivers value and stays true to the brand, it still feels authentic.
In audio, the progress is even more striking. Using tools like ElevenLabs, I now find it genuinely difficult to tell whether a voice was recorded by a real person or generated by AI. It used to be very obvious, but it’s no longer so. Furthermore, platforms like Suno are even generating full songs with AI, demonstrating the creative potential of this space. While AI still can’t fully replicate the emotional nuance of human storytelling, it’s incredibly close in certain contexts, such as podcast intros or voiceovers.
In the video, I don’t think we’re quite there yet. Viewers still want to see real people behind the brand. AI-generated avatars or synthetic presenters can feel impersonal or even off-putting. In this format, over-automation can actually hurt trust rather than help it. Video remains one of the formats where human presence matters most.
In images, there has also been significant progress. The latest image generation tools, including those from ChatGPT, have significantly improved in detail, structure and realism. That said, creating truly high-end visuals or design assets still requires a designer’s eye. Similar to video, AI-generated images are great for speed and experimentation, but harder to use when the goal is premium quality or emotionally resonant storytelling.
So yes, authenticity is absolutely possible at scale, but only when AI is used with care, intention and clear human oversight. It’s not about hiding AI, but about using it to support more relevant, consistent and thoughtful communication.
I believe AI enhances brand trust when it helps a company communicate with clarity, consistency and speed, especially in environments where multiple teams or tools are involved. If we take the example of our custom GPT app, it doesn’t just automate content creation. It acts as a quality layer that ensures everything we produce aligns with our tone, structure and brand voice. That kind of internal alignment leads to external consistency, which is one of the foundations of trust.
Where AI starts to undermine trust is when it’s used in ways that feel impersonal, overly generic, or disconnected from human expectations. As I mentioned earlier, I do not think we are there yet with video. People still want to see real people behind the brand. AI-generated avatars or synthetic presenters can feel off or even off-putting. In this case, over-automation can actually hurt trust rather than help it. The same logic applies to other areas where emotional nuance or authenticity is expected. If AI steps in too aggressively, without context or empathy, users pick up on it quickly and the experience can backfire.
For me, the difference lies in intent. If AI is used to support the user experience by helping teams respond faster, personalise better, or maintain quality, it builds trust. However, if it’s used purely for volume or to eliminate human oversight from the process, it risks damaging the relationship instead of enhancing it.
For me, ethical considerations with AI in marketing come down to common sense and responsibility, not just technical rules. I don’t believe we need to disclose every time we use AI, especially if the output is refined, edited, or built upon by a human. If something is 100 per cent AI-generated and used publicly as-is, then yes, it should probably be transparent. But if it’s part of a creative workflow, I don’t see a need to over-explain it.
Where I draw the line is using AI to build on someone else’s work without permission or credit. That’s not just unethical, it’s lazy. Just because AI can remix content doesn’t mean we should ignore the original creator behind it.
In general, I think the bigger issue is the temptation to overuse AI because it’s trendy. Ethical marketing is also about being intentional, not just using AI because it’s available, but asking whether it actually improves the outcome. I constantly ask myself if AI can genuinely help me solve the task at hand, and often the answer is no. Either the quality isn’t there, or the time spent tweaking the result outweighs the benefit.
So while ethics are important, I’d say the more relevant mindset is thoughtful adoption. Don’t get caught in the hype. Use AI where it truly adds value and skip it where it doesn’t. That, to me, is a much more sustainable and honest way to work.
Yes, AI has definitely influenced my creative process, but I would describe it more as an enhancement than a replacement.
Take content creation, for example. When working on content briefs, I usually combine manual keyword research with AI to generate a first draft. AI tools are quite good at determining the actual intent behind keywords, which is a great starting point. But the best results come when I restructure the brief myself to match the user journey and purpose of the article. Without that human layer, the output tends to be mediocre at best.
When it comes to publishing, I think it’s perfectly fine to use almost fully AI-generated content for informational or glossary-style articles, or anything that is factual and straightforward. But for opinion-driven pieces, thought leadership, or anything that carries brand voice, AI just isn’t enough. It lacks the subtlety and conviction that readers expect from a human perspective.
In ad design, my experience is mixed. I’ve tested AI-generated ads for product catalogues in e-commerce, and the quality can be surprisingly solid when you have thousands of SKUs and need scale. But for general brand ads like the ones we produce for Luigi’s Box, the outputs are mediocre. I don’t see AI replacing skilled UI designers anytime soon. The nuance, brand context and creative flair just aren’t there.
Where I’ve found real value is in campaign ideation. AI works well when I use it to expand my thinking. It helps me explore angles I already knew deep down but hadn’t considered at the moment. It’s like a creative partner that brings up ideas I might otherwise overlook. The key is knowing what to use and what to ignore, because the volume of suggestions can be overwhelming.
And that circles back to the core belief I hold – AI is not a replacement for human creativity. It is a tool that enhances it when used with intention and experience.
I think the role of human marketers is shifting, but not in a linear way. Right now, I see two types of marketers. There are those trying to become AI experts, investing time in prompt engineering and automation, and then there are those jumping on the hype, applying AI to every task just because it’s trendy. Honestly, it’s difficult to do both well. I personally don’t have enough time to constantly stay up to date with every new tool or capability. So I always ask a simple question before jumping into anything AI-related: can this tool actually help me with this task, or would I do it faster manually? A quick bit of research is often enough to answer that, but it takes experience to judge it properly.
AI is entirely dependent on the quality of your input. If your data is fragmented, inconsistent, or messy, AI won’t fix that. In fact, I think many marketers will soon hit that wall and start going back to basics, making sure their underlying data and processes are clean before they can get any real value from AI. That foundational work is going to be more important than ever.
Another thing I notice is the growing trend of AI specialists offering to automate anything for you. And although that’s technically possible, it is not just about automation. Marketers themselves need to understand what should be automated and why. If they do, chances are they’ll be able to implement at least part of it themselves. The real challenge is finding the time to pause, rethink processes and integrate AI meaningfully into your workflow — and that’s not easy when you already have a full plate.
Looking ahead, I think AI tools will become more advanced not just in capability, but in usability. Most marketers won’t need to build anything from scratch. Instead, they’ll be able to choose from a library of proven workflows and templates and simply say, “Yes, that’s the task I’ve been doing manually. Now I can automate it.” That’s when AI will become truly embedded in everyday marketing — not by replacing us, but by making us more efficient at what we already know how to do.
I don’t see any reason why personalisation and privacy couldn’t go hand in hand. In fact, with GDPR and similar regulations in place, I think we’re more privacy-focused as an industry than ever before. Yes, we’ve lost access to a lot of data that used to make personalisation easier, but that doesn’t mean it’s impossible. It just means we need to be more thoughtful about how we collect and use data.
Personally, I rely mostly on first-party data — things like website behaviour, CRM activity and customer conversations or call notes. That kind of data is not only more accurate but also ethically sound because we have clear ownership and consent. AI can definitely help in making sense of it, but only when we’re working with data we actually control.
For companies just starting out, it’s admittedly a lot harder. Without a solid base of first-party data, personalisation efforts can feel generic or forced. But that’s where being transparent comes into play. If people understand what data you’re collecting and how it’s used to improve their experience, they’re far more likely to opt in and trust your brand.
Let’s also not forget that privacy isn’t just about regulations — it’s also part of the brand experience. Users need to feel safe when they engage with you. Being too aggressive with personalisation or misusing AI can quickly undermine that trust. One important rule I follow is to segment users for analysis and personalisation, but avoid ever revealing individual identities, especially when using AI to train models or generate content.
Also, attribution has become more difficult than ever. With platforms like Google increasingly limiting tracking, it’s no longer realistic to try to measure everything. I’ve actually moved away from obsessing over complete attribution. Instead, I focus on doing meaningful work that I know contributes to long-term impact, even if I can’t track every click or view. Since making that shift, my mental health has improved significantly. I’m more focused, less frustrated and more confident that consistent effort builds results over time.
So for me, the balance is about working with what you have, being transparent and building personalisation around consent, not assumption. When done right, privacy doesn’t limit personalisation — it makes it more sustainable and trustworthy.
I’ve faced quite a few challenges when integrating AI into my marketing stack and workflows. Some of them I’ve already mentioned, like mediocre outputs, unreliable automation, or poor data quality. But in practice, the real challenge often lies in the gap between what AI promises and what it can actually deliver at scale.
One of the biggest limitations I’ve seen is around data. AI is only as good as the data you feed it, and if your inputs are messy, outdated, or incomplete, the results won’t be reliable. The same goes for tools that claim full automation but lack the necessary integrations or flexibility. Sometimes I want to automate a task that genuinely takes most of my time, but the tool either doesn’t support the system I use or can’t complete the workflow from A to Z.
One concrete example where AI actually worked well is with HubSpot. They recently rolled out a feature where you can type a natural language prompt, and it creates a workflow for you. It’s surprisingly helpful and saves a lot of time, but even there, it’s not perfect.
First, you need to have clean, high-quality data already in your CRM. Second, the prompt has to be well thought out. You really need to consider all the conditions, branching logic and segmentation rules; otherwise, it won’t do what you expect. And third, you usually have to go in afterwards and manually fix errors or tweak the logic.
So yes, AI speeds up the process, but it still needs a lot of human input. Without that, it could easily lead to failure. It’s not a plug-and-play solution. You need to understand the system, the logic and the audience. Otherwise, it becomes more of a risk than a time-saver.
To be honest, I’m a bit concerned about where things are heading based on what I’ve seen recently. AI is evolving fast, and I think we’re getting very close to a point where a lot of the work traditionally handled by agencies — like building websites, landing pages, writing copy, designing visuals, or managing social media — can now be done in-house, faster and more affordably, with AI tools. That’s going to be a big shift.
Creative agencies might still have a place, especially when it comes to brand strategy and high-impact campaigns, but for smaller companies, an AI-driven in-house setup will often be the better, more agile option. Roles like web editors, proofreaders, translators and other support functions are already being replaced or heavily reduced. That trend will only continue.
One interesting shift I see coming is the normalisation of “vibe coding” and “vibe marketing.” Currently, it’s trendy, but in a few years, this will simply be the standard way of doing things. You’ll sketch your idea in natural language or visuals, and AI will bring it to life. But with that accessibility comes a downside. The internet will be flooded with similar-looking and sounding content.
In that environment, the winners will be the ones getting copied, not the ones doing the copying. Relying on best practices or just mimicking competitors won’t be enough anymore. There will be more players in every niche, and the barrier to entry will be lower than ever.
So I believe only companies with a strong brand identity and a personal, human touch will stand out. That uniqueness, whether it’s in voice, values, or experience, will matter more than ever. AI will level the playing field, but it won’t replace the need to stand out.
When it comes to measuring the ROI of AI-powered marketing initiatives, I try to stay grounded. My first question is always simple: what traditional alternative could I do instead of this AI-driven campaign, and would it give me a better outcome? A lot of AI initiatives sound exciting, but when you look at the opportunity cost, what else you’re not doing because of it, the justification often isn’t strong enough.
That’s why I measure AI campaigns mainly through two lenses: effort versus value. I look at micro-conversions like traffic, clicks, or engagement metrics, but I also factor in what else I had to postpone or deprioritise to run the initiative. For me, prioritisation is everything. If the AI-powered idea doesn’t clearly outperform what I could have done with traditional methods, it’s not worth it.
That said, I don’t think the measurement framework itself is different. I still look at the same KPIs, such as traffic, conversions, leads, or revenue, but the variables are different. For example, if one content writer can produce a human-written article in a day, and the same person can produce 15 AI-assisted articles with similar effort, the ROI analysis gets interesting. Sure, the 15 articles might bring more traffic. But do they bring more signups, purchases, or meaningful interactions than one high-quality article? In many cases, I’m not convinced they do.
So I try not to get caught in the volume game. ROI isn’t just about doing more with less. It’s about doing the right things better. AI gives us speed and scale, but the true value still comes from intent, context and execution.
One thing I see marketers getting wrong about AI is the urge to automate everything blindly. There’s this mindset that if something can be automated, it should be, and that’s just not true. It’s not about AI versus not-AI. It’s about finding the right balance between efficiency and thoughtfulness.
AI is a great enabler, but without context, strategy and human judgment, it often leads to mediocre results or even harm to the brand. Just because a tool promises end-to-end automation doesn’t mean it will deliver meaningful outcomes. I’ve seen teams automate content creation, ad targeting and even customer segmentation without really understanding what’s happening under the hood. That approach usually backfires.
Instead, marketers should ask: where can AI actually improve my process without compromising quality? Can it help me think better, act faster, or support decisions with data? That’s a far better mindset than simply trying to replace yourself.
A good example from Luigi’s Box is how we use AI for product discovery. Our models use a mix of user signals like search queries, click behaviour, past purchases and session context to personalise results in real time. But we didn’t just plug in AI and let it run. There was a lot of iteration, testing and manual oversight to get the logic right and make sure it actually improved user experience. The AI supports the customer journey, but human guidance shapes how it works.
Another misconception is thinking that AI is a one-time setup. In reality, it requires iteration, refinement and constant supervision. You can’t just hand over your workflows to AI and expect consistent results. You need to understand the logic behind it, monitor outcomes and continue to provide high-quality input.
The focus should be on amplifying creativity, freeing up time for strategy, and ensuring that the outputs still reflect the brand and its values. When AI supports your thinking, not replaces it, that’s when it works best. Long term, the marketers who succeed will be those who learn to collaborate with AI, not compete with it. It’s a partnership, not a shortcut.
If I could give one piece of advice to marketing teams just starting to explore AI, it would be this: challenge yourselves. Not just by learning what AI tools can do, but by actively thinking about how you can use them to produce more and better output and enhance the value you bring to the business.
Don’t approach AI as a shiny new toy. Instead, approach it as a way to evolve your capabilities. Ask yourselves what tasks are slowing you down, what areas of your workflow are repetitive or inefficient, and where your output could be more strategic or impactful. Then see where AI might help you scale, improve, or accelerate those parts.
This goes beyond just efficiency. It’s about making your work more valuable to the business. If you can use AI to turn a one-week project into a two-day project without sacrificing quality, or if you can support more channels or campaigns with the same resources, your position in the company becomes stronger. You’re no longer just executing tasks. You’re bringing leverage. That’s what every business wants right now.
But be honest with yourself. Not every task benefits from AI, and not every tool will fit your use case. The goal is not to automate everything blindly. It’s to understand where AI can help you be faster, sharper, or more creative. When you start thinking this way, you stop being just a user of tools and start becoming someone who drives the business forward.
That mindset will not only make you more productive but also more future-proof.
