9 ways growth is different in AI companies

It’s kind of the same game... but the rules have changed.

A year ago, I would’ve laughed at the idea of running growth at a $130M ARR company without touching activation.

Or that 90% of my time would be spent on big bets instead of optimizations.

Or that our biggest marketing wins would come from our founder's LinkedIn posts, not our paid marketing strategy.

But here I am at Lovable, watching the rules of growth turn upside down in real time.

If you're a growth operator at an AI company and feeling like you're making sh!t up as you go, you're not alone. We all are.

---

This post is sponsored by one of my new favorite products - Mobbin! It’s the most comprehensive UI/UX design reference library for mobile and web apps. No more spinning up fake accounts just to spy on product flows. Now I just open Mobbin. I’m obsessed.

---

Here are nine differences I see every day doing growth at an AI company compared to traditional tech software.

1. PMF treadmill

In most companies, PMF locks in around $1–2M ARR - that’s when the switch flips to scaling, and growth, marketing, and sales take the wheel. By $100M, you’re usually pushing into second-horizon bets or spinning up a multi-product strategy.

AI-native companies don’t follow that script. At Lovable, we’re well over $100M ARR and still re-earning PMF every month.

In AI, nothing stands still and categories evolve at lightning speed: tech shifts overnight while customer expectations reset on monthly basis. Don’t even get me started on LLM capabilities that evolve at breakneck speed. Competitors are on the same PMF treadmill and running just as fast. And yes, we are all racing each other.

That means the product is in a constant fight to re-prove PMF.

In non-AI companies, you work on a more stable foundation, but in AI, that foundation resets itself every few weeks. That makes growth constantly secondary to what is being worked on by core.

2. Activation happens in the prompt box

For years, I preached about activation.

I’ve claimed that it belongs to growth. That poor activation kills retention. That without it, monetization stalls and virality never lights up.

But in my new role, I don’t even work on activation… Gasp.

Before, there were surveys, setup flows, feature toggles, onboarding screens… lots of surface area where users could get lost, and growth could step in to guide them toward value.

But in AI-native companies, you have literally one interaction point: the prompt box.

For Lovable it’s this. This is it! I mean… not much to work on (besides core agent experience, obviously)… right!??

Think of ChatGPT - the user types, it responds. The user either gets value or they don’t.

That interaction is the activation, and it’s entirely driven by the core product. Growth can support with education and inspiration (helping users imagine what to ask and what’s possible), but the crux of activation belongs to core product.

That’s a good thing. It forces product teams to own activation and value delivery directly, instead of shipping features and expecting growth to work on activation later.

3. More big bets, less optimizations

In every job that I’ve had, I dreamt of setting up new growth loops or unlocking new channels.

Yet, what was I doing 90% of the time? Optimizations.

Optimizing activation, free-to-paid conversion, renewals… you name it.

At Lovable, I’ve completely flipped the ratio, and most of my time is spent on building the foundations. AI-native companies are inherently young, so there are many building blocks to put in place before there’s anything meaningful to optimize.

For example, we launched free collaboration, a huge driver of virality. But have we optimized the flows? Not really. Instead, we focused on getting the next big block in place. Again, same story: we created it, but left optimizations for later.

There’s so much natural curiosity and demand around AI right now that optimization simply isn’t where the biggest impact lies.

I’ve never spent so much time in the big bets territory in all my career, even at companies with lower revenue than Lovable. When I joined SurveyMonkey at exactly the same revenue level, all I did was optimize existing systems to reach the next $200M. But the ceiling for some of these AI companies is so much higher that the foundational work still makes more sense than incremental improvements.

4. The growth marketing playbook collapsed

For years, the marketing playbook was predictable: start with SEO for scalable organic traffic, layer in user-generated content, add paid marketing, then experiment with social channels.

Ten years ago, that was enough to win, but not anymore.

All the attention has shifted to the creator economy.

Eyeballs are now on YouTube, TikTok, Instagram, newsletters, podcasts, and LinkedIn. (They are also on AI too, like ChatGPT, but there’s minimal distribution there…).

Even in B2B, the mix now looks almost like consumer: Influencers, creators, and video.

Take YouTube: 10 years ago, it barely mattered for B2B. Today, if you’re not building there, I don’t know what you’re doing.

The source of content has changed too. The goal is no longer to produce everything in-house, but to empower creators to generate content about you, and then repurpose it across paid and organic.

5. Marketing can’t keep up with shipping velocity

AI products ship at a pace most marketing teams were never designed for.

At Lovable, we often ship multiple times a week. In most companies I’ve worked in, we launched quarterly, maybe monthly. Marketing had plenty of time to prepare, build narratives, and design campaigns.

But what happens when the product is shipping every other day?

You can’t put all hands on deck for every update.

Slow down shipping to give marketing time, and you lose velocity. Push updates without marketing support, and you lose awareness.

The solution isn’t obvious yet. What I do know is that marketing has to move closer to product and get comfortable working at this speed because product velocity can’t slow down. It’s on Marketing to adapt.

6. Brand became a product job

Software supply has exploded recently.

Everyone can launch something (that’s exactly what we’re enabling at Lovable).

But in a crowded market, how do you stand out? Through brand.

It’s not like brand didn’t matter before - it did. But it used to be a marketing exercise, and you wouldn’t connect it to much beyond campaigns and positioning.

Now, with the abundance of software, brand needs to be channeled through every single product interaction. It became the job of designers, engineers, R&D teams, and everyone else to embed brand values into the user experience.

It’s similar to a traditional PLG motion, where every user becomes a marketing ‘agent’. And that experience is stronger the more you infuse it with "love marks": moments that communicate your intent, message, and how you want people to feel.

It’s not the utility-based features but the moments that create human connection within the product, like an unexpected reaction, a playful animation, or a confirmation written in a warm tone. In Lovable, we do this by ensuring every single experience is “lovable”.

7. Founder-led social is the hottest ‘new’ channel

Ten years ago, the organic strategy that scaled was SEO. Today, it’s founder-led communications.

In B2B, that plays out on LinkedIn and X. In B2C, it’s Instagram and TikTok. But the principle is the same: direct communication from leaders outperforms polished corporate messaging.

A founder or executive with an authentic voice can now drive reach at a scale PR never could.

At Lovable, Anton’s account generates millions of impressions for us. That visibility drives awareness and shapes people’s perception of the company.

Every CEO and executive should be thinking about how they communicate to the market. And the more people inside the company who do it, the better.

The challenge is that large companies often restrict employees, worried about losing control of the message. Employees themselves usually aren’t sure what they’re allowed to share. But if you can build enough trust to let both founders and employees communicate openly, their voices become your best growth strategy.

At the end of the day, it’s the humans inside the company who carry the brand further than any campaign ever will.

8. Product-led loops matter more than ever

Every channel is a knife fight out there. Paid gets pricier, algorithms shift, influencers churn.

The only channel you truly own is your product - and it has to carry more of the growth load than ever. Word of mouth, UGC, referrals, collaboration loops - whatever the mechanic, bake it into the product itself. Make the product-led growth a native experience, not an afterthought, and push it until it compounds.

9. Growth ships features now

I’ve always said core builds features and growth distributes them.

That division doesn’t hold in AI-native companies.

The ground shifts too fast, and distribution alone isn’t enough. My growth team isn’t just optimizing funnels - we’re shipping features. Sometimes those features are the growth loop itself. Sometimes they’re table stakes to keep pace with user expectations (example, growth shipped voice mode for Lovable). Either way, growth isn’t sitting on the sidelines waiting for core anymore. We’re in the product trenches, building alongside.

Same game, new rules

Doing growth in AI-native companies feels like playing the same game on a new board. The pieces look familiar, but the rules don’t apply.

With distribution changing monthly, technology evolving weekly, and customer expectations resetting daily, we’re all figuring this out together, in real time.

There’s no fixed formula, only constant adjustment. And honestly, that’s the best part of the job.

Edited with the help of Diana Bernardo.