When Elena Verna, Head of Growth at Lovable, took the stage at SaaStr AI, she’d just hit her one-year anniversary at the company. She walked through the question many B2B leaders are wrestling with: when AI writes 80%+ of your code and anyone can vibe code your feature set in an afternoon, what’s actually left to compete on?
Here’s what she said, and the takeaways worth stealing.
1. Lovable Is at $400M ARR With Fewer Than 200 People
Lovable shared in February that it crossed $400M in ARR. The team is still just shy of 200 people. That’s north of $2M in ARR per employee, and they did it in a category that barely existed two years ago.
The headcount discipline isn’t an accident. Verna calls the structure “product engineering” and says they deliberately reject the old ratio of one PM to seven engineers to a designer to a marketer. Everyone does some IC work. Everyone ships. The result is a revenue-per-head number that simply wasn’t possible in the pre-AI org chart.
Even approaching half a billion in revenue, Verna says Lovable is still “on the product market fit treadmill.” The category is moving so fast on both the technology and the customer side that they feel like they have to recapture PMF every month. Scale didn’t let them slow down. It raised the stakes on velocity.
2. Feature Differentiation Is No Longer a Durable Moat
In AI native organizations, 80%+ of the code is now written by AI. When the cost of building collapses, feature parity stops being a years-long engineering effort and becomes a weekend. You might be ahead for a month, two months, six months. Then everyone catches up.
For the last 15 years, B2B companies leaned on feature differentiation as the moat. We won because we had the better product, the better engineers, the better product visionaries. Verna’s point is that you can still get a feature lead, it’s just short-lived now, and you cannot build a predictable growth engine on top of something competitors can clone in weeks.
The moats that still hold:
- Hardware. Still genuinely hard to build.
- Network effects. Always hard to create, and once you have them, they keep compounding.
- Data. Especially proprietary data competitors can’t replicate.
- Security and compliance. Slow, expensive, and worth investing in for exactly that reason.
- Brand. “Brand is back, baby.” When everyone can build the product, the relationship with the customer is what’s left.
Note what’s missing from that list: SEO and SEM. More on that below.
3. The New Career Flex Is the High-Powered IC, Not the VP
Verna ran a growth team of a couple hundred people at Dropbox, where growth was bigger than the entire marketing org. At Lovable she went the other direction on purpose. She fired herself out of the marketing job, then handed off the growth lead role, and went back to being an individual contributor.
Her read on the next decade of careers: the flex is no longer climbing toward the fancy VP title. It’s becoming the high-powered IC who can do, with a stack of agents, what used to take dozens of people.
She’s blunt about why so many leaders are unhappy. The reward for being a great IC has always been a promotion into management, which is a completely different job that most people were never built for. Her advice to founders trying to find their own version of this person: go reach out to the leaders who are quietly miserable in coordination roles and ask if they want to build again. A surprising number are saying yes, because they see it as a way to fall back in love with the work.
4. Flat Org, No Titles, and Everyone Ships to Production
Lovable runs with no internal titles. Not as a culture gimmick, but because everyone is expected to do real building, including the people at the top.
A few mechanics that make the velocity real:
- A
#shippedSlack channel where the day’s production releases pile up. Multiple ships a day, not a sprint cycle. - A
#feedbackchannel where ideas surface and, if the team agrees, go live in 24 hours. - An operating principle: if you can convince one other person it’s a good idea, go build it.
That last one only works because everyone holds enough agency that getting even one person to agree isn’t automatic. Verna’s favorite example on herself: she vibe coded a full redesign of the enterprise pricing page, opened a PR, and was ready to ship. A 20-year-old engineer told her to go get a design sign-off first. No hierarchy override. Everyone pushes back on everyone, regardless of who the idea came from.
Compare that to her old life at the big companies, where a leader’s idea got waved through with little real pushback and took months to ship anyway. The speed is the obvious part. The deeper change is that information flows all the way down, so anyone in the org can challenge a decision on the merits.
5. Freemium Is More Important Than Ever. Treat It as Marketing Budget.
Verna’s contrarian tactic: freemium has never mattered more than it does now, and the instinct to gate your high-cost AI features is exactly backwards.
Yes, the bill is scary. Giving away AI-heavy product is expensive in a way that giving away seats never was. Her framing is to stop treating that cost as COGS and start treating it as marketing spend. It’s the only reliable way to get into customers’ hands and change their habits.
The proof point: Lovable announced a partnership giving every LinkedIn Premium member ad-free access to the product. Conversion rates from that cohort to paid are running in the double digits. The lesson she keeps repeating is to ungate, not gate, to win this market.
6. Context Is the Real Moat for the Rest of Us
An agent will produce average output unless you feed it enough context. Trained on the open web alone, your future AI double is just the average intelligence of everyone. The differentiator is your context: your call recordings, your ideas, your brainstorms, the way you actually think and decide.
Verna’s advice is to start capturing that now, before you need it. When the time comes to spin up agentic versions of yourself to run workflows, the ones built on your proprietary thinking will do your work. The ones built on nothing will do everyone’s work, badly.
This is the operator-level version of the data moat. Most people have no proprietary context captured. The ones who start now will have a real edge in 12 months.
7. SEO and SEM Are Now Table Stakes, Not Winning Moves
SEO used to be a reason a company won its market. Now it’s something everyone has to do, and it won’t be why you win. Same with paid. You’ll probably run it, but it’s the cost of existing, not the source of the advantage.
This reframe matters because a lot of B2B teams are still organizing growth around channels that have quietly demoted themselves from moat to maintenance.
8. Buy vs Build Isn’t Binary, Even at a Vibe Coding Company
You’d expect the company built on “build it yourself” to be 100% build internally. It isn’t.
Lovable uses Linear for project management because it’s too deep in the features they need. They use Slack. They use Granola for meeting recordings. They evaluate every purchase hard, but they don’t pretend the answer is always build.
Where they do build: their own internal admin tool (named “Woof”) that lets support grant credits, create coupon codes, and add features on the spot in five minutes. Their own CRM, which the sales team actually loves. The answer, in Verna’s words, lives in the middle, set by the complexity of maintenance versus the feature richness the tool actually needs.
For everyone watching the buy vs build debate play out, that’s the honest position. Build the satellite tools and the workflows nobody else will ever build for you. Buy the deep, well-built systems where someone has a decade head start.
Top Mistakes Elena Shared
She was direct about what she got wrong. The short list:
- Going into management in the first place. She calls herself a “mediocre manager” and admits her superpower was always craft, not coordination. Years got spent running large teams when the IC seat was where she did her best work.
- Vibe coding chaotically instead of starting with a spec. Her own style is “this thing sucks, do this, no, this looks wrong,” and she points to a clean, structured prompt as the better way to start. Skip the chaos at the front and the agent gets you there faster.
- Trusting her own first idea too much. She now prompts the agent to build her version plus two better options, because her original idea is often the worst of the three. The fix was getting out of her own way.
- Trying to ship without the right check. She vibe coded a full enterprise pricing page redesign and was ready to push it live until a 20-year-old engineer made her get design sign-off first. Fast does not mean skip the one review that matters.
- Slipping back into pre-AI habits. Even now she catches herself operating the old way and has to stop and re-think the problem for an AI native team. The mindset shift is a transition, not a switch.
