What's It Like to Work at an AI Native Company?
Field Notes from a Lovable Employee
A Field Report from Inside Lovable
🙍♀️ Author: Elena Verna
🥷 Translated by: NCon
Editor's note: Everyone's talking about how AI will reshape our companies. The article we're featuring today is a first-hand account from an employee at Lovable, a representative AI Native startup that's already hit $500 million ARR.
What's interesting about this piece is that it captures both the exhilarating and the bewildering — even outright strange — aspects of working at an AI Native company. Many of you will likely find something to relate to.
And for those who haven't yet worked at an AI Native company, this is a rare glimpse inside.
Working at Lovable Does Feel Different
By now, I've worked at or advised quite a few tech companies. Big ones, small ones, public ones, fast-growing ones, slow-moving ones, and everything in between.
But I also belong to a generation that came of age in the previous tech era. Most of my instincts about "how companies should work" were formed before AI: managers manage, everyone has a clearly defined job, important decisions get made in meetings, information flows through the org chart, and seniority usually means managing more people.
After a year-plus at Lovable, I keep running into things that clash with those instincts.
This is the first company I've worked at that I'd call "AI Native." Let me be clear — I have exactly one data point here, so this isn't a manifesto for how AI Native companies "should" operate.
I'm not even sure which characteristics come from being AI Native, which from Swedish culture, and which are just Lovable being Lovable.
But something is genuinely different.
I'm not talking about "we use AI a lot." Most discussions about AI at work still focus on the operational layer: writing emails faster, generating code, summarizing meetings, making slides, researching customers, automating support.
These are productivity gains. But if a company just slaps them onto the same old organizational model, the end result still feels somewhat flat.
What actually feels foreign is this: when you assume AI will be involved in everything, the organization itself starts to change. What work looks like, how people collaborate, who makes decisions, how information moves, even what it means to do a good job — it all shifts.
So consider what follows field notes from someone whose "operating system" was installed pre-AI.
Some things are great, some are hard, and some still feel... weird.
What I Like
We Don't Have Layered Titles
Yes, I was on board with this from the start. We have ICs, Leads, and Heads, but no further title granularity.
This was almost a necessity. We're growing so fast that we didn't want to end up in the familiar trap where someone excellent at their director-level job today is still just as excellent next year, but the company has grown 10x and their scope now looks like what other companies would call a "manager" role.
Then we'd have to awkwardly approach them about a title change. It's not a demotion. But it is. Except it isn't, because the company just got bigger. But it absolutely feels like one to the person.
Why do we do this to people? It's basically a motivation destruction machine.
I hadn't fully appreciated how much organizational machinery and office politics are attached to layered titles. Titles aren't just labels for jobs; they represent status, access, information, even permission to act.
Depending on where you sit in a company's org chart, you may not even realize how real this is.
You get promoted, and suddenly you're pulled into different meetings, invited to different Slack channels, cc'd on different decisions, and shown information that — apparently — 24 hours ago, you weren't senior enough to know.
By contrast, Lovable's structure is so flat it's almost disorienting. People opt into discussions and opt out on their own. Unless there's a good reason, nearly all Slack channels are public. Information doesn't just flow up and down management chains.
Strip away a few traditional layers and you start to wonder why they existed in the first place.
One thing I still haven't fully figured out: what does this do to management itself?
Growing up, I absorbed a long list of things managers were supposed to do: relay information, coordinate work, approve decisions, assign responsibility. Here, we don't seem to need those things as much.
Managers still exist — we call them Leads and Heads. But their work seems less about controlling the flow of work and more about helping people understand context, raising quality standards, developing people, and making sure a group of highly autonomous individuals is running very, very fast in roughly the same direction.
I suspect management is one of the jobs AI Native companies will need to rethink most. And again: this doesn't mean managers will disappear.
It's just that a significant portion of what we used to call "management" may have been moving information and decisions through layers that are no longer necessary.
Senior People Are Happy Being ICs
At Lovable, there's a surprising number of senior people working as ICs.
Some of them managed dozens of people in their previous roles, and now they're just hands-on, shipping projects and tasks directly. Folks, we have entered the era of the High-Impact IC — the HI-C!
I've seen this elsewhere, of course, but it was always shadowed by something. Here, staying an IC or returning to IC work no longer implies your career stalled out halfway.
Partly, it's because AI capabilities expand what one person can do. Now, a strong individual can handle research, building, analysis, writing, prototyping, shipping, and iterating — work that used to require a much larger team.
But partly, it's just a stupid system naturally rebalancing itself. That system keeps pushing people into management even when they don't want to go.
Even when the new role would completely prevent them from continuing to create the value that made them stand out in the first place.
It's so sensible that I don't think it'll stay limited to AI Native companies. I did my own research on this and found that 71% of people, including managers, want to spend at least half their time on IC work[1]. So I think it's only a matter of time.
Everyone Is Deeply Suspicious of Meetings
I have zero complaints about this. I spend less than 20% of my time in meetings, which would have been unthinkable at some places I've worked.
Clearly, it's not because we're accomplishing less, or because cross-team collaboration has dropped. It's just that meetings aren't the default way to share information here.
Partly, our AI tools make sharing context much easier. If you can pull up the company's shared knowledge base anytime and see what's happening, including why a decision was made, you don't actually need to attend that meeting.
But the bigger reason is that initiative and autonomy are strongly encouraged up and down the company. You don't need a "license" to do something, and you don't have to route every decision up and down the org chart for approval.
We want to move fast. And when you move fast enough to course-correct quickly, the cost of doing something wrong drops dramatically. In fact, the cost of being wrong is usually lower than the cost of endless meetings to reach a perfect decision — which doesn't exist anyway.
So yes, we get things wrong sometimes. But that's learning. It produces valuable context, we adjust quickly, and we keep moving.
What's Hard for Me
The Pace Is Insane
I often say I've "raw-dogged" more than a year at Lovable — and I'm only half-joking.
People in tech, especially in AI, get excited about the speed of shipping and innovation. But it's genuinely exhausting.
Everyone's scrambling to capture territory — to ride this AI wave, to lock in users before they're captured by competing products; and above all, because everything could change overnight.
If a few major labs release new models, the core capabilities of our product might look completely different. Or a competitor might suddenly be able to do things that were previously impossible.
Much of Lovable's design is meant to create an extremely fluid work environment so we can adapt to all this. But teams still get hit head-on by the chaos.
A few months ago, I took a week off. When I came back, I genuinely felt like I was onboarding at a brand new company.
Almost everything had changed, and I had to sprint to catch up with everyone else... who were all sprinting at full speed themselves.
Role Boundaries Are Blurry
At Lovable, there's almost no boundary between different jobs. People jump into things that technically aren't theirs, start projects, build things, fix things, and occasionally work on the same thing in different directions simultaneously.
I've had too many moments where I assumed something was my responsibility, only to find someone else had already taken it in another direction. And of course, I've done the same thing myself — seeing a problem I wanted to solve and walking into another team's territory.
For much of my career, this has actually been my default way of working. So you might think I'd be 100% thrilled about this.
At many companies, especially large ones, role definition and boundary-setting is a major problem in itself. And it's not just petty turf wars.
Sometimes people just default to "not my job": if you get criticized for fixing a bug that "belongs" to another team, you're not going to go out of your way to fix bugs anymore, right?
So part of me genuinely loves that at Lovable, everyone seems capable of doing everything. But it's also draining. There's a reason companies spend so much time clarifying who's responsible for what.
I'm grateful that Lovable's culture is encouraging and forgiving, and people don't hold grudges. But when someone suddenly runs into my area and appears to be doing my job, I do find myself getting territorial.
Then again, as the saying goes, two heads are better than one. So I've had to learn to loosen my grip and accept that everyone is somewhat doing my job, and I'm somewhat doing everyone else's.
Honestly, I still don't know exactly where the line is. Sometimes "high initiative" is just a generous term for duplicate work, unclear ownership, or someone stepping into someone else's space.
Sometimes chaos produces better results. Sometimes it's just chaos.
We're in Constant Reorganization
At Lovable, the only constant is change. Teams are always shifting: responsibilities move, people switch to different areas, new teams emerge, old ones disappear. Arrangements that made complete sense three months ago are now completely mismatched.
Here, reorganization isn't treated as a special "event." It's more like a steady state.
At some companies I've worked, constantly changing plans was mainly a symptom of leadership confusion. At others, merely saying the word "reorganization" was enough to derail a quarter's worth of work.
Leadership announces a change, actual work stops, and everyone scrambles to interpret what it means for their projects, their roles, their teams, their life plans.
What's different at Lovable is that everyone assumes the structure will change. The environment we're operating in is updating at a crazy pace, and combined with our extreme growth, operational phases that would take years at other companies get compressed into months here.
The structure that fit the company six months ago probably shouldn't still be the one today. So people are less likely to interpret reorganization as "something went wrong" and more likely to see it as "we learned something new."
But that doesn't make it easy. Constant change still makes the work harder. Running fast is already hard. Running fast when the stakes are high is incredibly stressful.
If you're moving at high speed with high risk, and your team, scope, priorities, and sometimes even the very ground beneath you are shifting... this is absolutely not a work style that suits everyone.
So the change isn't painless. It's just easier to handle when you know it's coming.
The Weird Parts
There's a Whole Bunch of "Cowboys"
Lovable has a lot of what I'd call "cowboys"[2]. They're high-initiative generalists who see something that needs doing, yell "yeehaw!", and charge straight at it.
They don't wait for someone else to make a plan for them. And if there is a plan, they may not pay much attention to it. They find uncertainty exhilarating; even having to cobble something together at the last minute doesn't faze them.
At every stage of the company's expansion, Lovable has maintained a high ratio of cowboys because... we kind of have to. We're running on the product-market fit treadmill[3], and innovation doesn't stop, can't stop.
This is... mostly great. I like to consider myself a cowboy too (not why I moved to Nashville, but it does fit). And it's common at startups.
But it's less common at companies with $500M+ ARR that are scaling up. That stage usually belongs to "farmers": putting down roots, building systems, making things grow in predictable, plannable cycles.
As our systems evolve and we increasingly need reliable infrastructure, we'll have to figure out how to balance these two.
People Are Actively Trying to Automate Their Own Jobs — and They're Excited About It
One of Lovable's stranger features is that someone joining and immediately asking "how much of my own job can I automate with AI?" is completely normal. And they genuinely mean it, happily.
People don't typically cling to tasks just because they happen to be written into their current job description.
For those of us who grew up with a different workplace mentality, this seems completely backwards. We were taught that job security comes from being irreplaceable.
Most of us were trained to keep accumulating more responsibilities and doing them very well. You become the person who owns the process, knows how the system works, produces the reports, answers the questions, runs the workflow.
So watching this happen is also instructive. If you assume the next sentence is "...and then we don't need you anymore," then "automating your own job" sounds terrifying.
But here, it's more like: "...so you can figure out what your job should become next."
I suspect this fear is much heavier in traditional organizations, where AI capabilities are often framed as a threat to job survival.
But at Lovable, the company expects people to make the most of this new technology. So if you can prove you can automate a set of complex or difficult tasks, that actually proves your value. And there's always more to do after that.
Our team is absolutely growing, not shrinking. (We'll be approaching 500 people by year-end... I joined when the whole company was 20. What?!)
This environment makes me grateful for my growth background. I've long been used to jumping in, solving a problem, handing it off, and chasing the next value-creating opportunity. But I know this isn't normal.
AI Agents Are Actual Colleagues... With "Parents"
Yes, this one still feels pretty weird to me.
We have agents everywhere: marketing agents, analytics agents, codebase agents, a general "company brain," and an increasing number of small tools people build around their own work. (You should see our GitHub agents swarming every PR!)
At this point, on a normal workday, I interact with AI agents almost as often as I do with actual humans. Is that kind of crazy??
What's also interesting is how we manage them: Lovable has a role called "agent parent." This person is responsible for the quality of an agent's output.
They teach it, update it, correct it when it starts giving bad answers — basically making sure it stays useful. The "parent" is usually whoever already had most of the relevant knowledge in their head and was having to answer the same questions for everyone else over and over.
So, okay, I'll admit it: this does feel a bit dystopian.
But watching it actually work, it's far less dramatic than it sounds. If you've answered the same question 50 times, you definitely don't want to answer it a 51st.
In fact, by that point, you've probably already documented the answer and context somewhere; the organization just never had a good way to surface that information and get it to the people who need it.
From what I've observed, this structure makes people's perspective and expertise more valuable, because literally everyone in the company can benefit from it. If everyone depends on this information, it better be right.
The Only Criterion Is: "Is This Lovable Enough?"
One phrase I hear constantly at the company, in various forms, is some version of "is this lovable enough?"
As a acceptance criterion, honestly, it's not very reliable. There's no checkbox on a list where you can tick "is this lovable enough?" So, see the pattern? It can feel a bit chaotic.
My sense is that it means you have to rely more on judgment. Everyone is more or less defining what "lovable" means according to their own taste; which means you have to find people with good taste.
And "actually lovable" is a higher bar than "works." Something can function perfectly, solve the problem it was meant to solve, and still not feel lovable.
So this standard forces us past "functionally correct" to ask harder-to-define things: does it spark joy? Is it intuitive? Are you excited to use it? Would you be proud to ship it?
No one can pull up metrics to answer these questions... which is exactly why judgment matters so much.
Right now, everyone in tech seems obsessed with the concept of "taste," so I won't belabor it. But when everyone is producing more, I can already see companies increasingly depending on people who have this magical "judgment."
This Is Not a Playbook
The strange paradox is this: the more AI handles the mechanical parts of work, the more companies seem to depend on human judgment.
When information is everywhere, execution is cheap, and everyone's capabilities are massively amplified, knowing what to do, when to step in, what counts as good, and when something is actually done becomes extraordinarily important.
"Let's have fewer meetings!" sounds great until no one knows what's happening. "Let's give everyone autonomy!" sounds cool until you realize everyone is autonomously running in different directions.
"Let's get rid of layered titles!" sounds nice until you realize the entire status system and decision-making machinery still runs on them. Whatever this is, it looks more like a whole system working together.
And I'm absolutely not saying other companies should work this way. I have exactly one data point, so please treat this as field notes, not scientific law.
That's partly why I wanted to write this down now. I'm old enough to remember how normal the old way of working seemed, before all this:
Managers held the context. Jobs had clear boundaries. Seniority meant managing more people. Important decisions happened in meetings. Software was something most employees couldn't just make for themselves.
None of this seemed particularly strange at the time. But I have a feeling some of it is about to seem strange — maybe already does.
Recommended Reading
Join Our Membership Group
