Every CEO I talk to is running the same loop in their head: am I moving on AI fast enough, or is my competitor already ahead?
That FOMO is the real story right now - not the technology. The tools work. The hard part is getting a real organization, with real people and real fear, to actually adopt them. This is a guide for that part.
Start with what changed. For years we all believed one thing: if you want more output, you need more people. Budgets, investors, HR - everyone stood on that side of the equation. AI broke it. The leverage is now big enough that a large chunk of the work just gets automated, so output and headcount come apart. That's freeing and unsettling at the same time.
And it throws everyone into mild panic. Three groups, three different flavors of it.
Investors start questioning how much money a company actually needs to grow - there's AI now, right? Boards, working closely with those investors, start challenging how every single team operates. We see it most often at the IT level: the CIO walks in and says, "My board wants us to ship more software, with a smaller team, and stop using external vendors - because the board feels AI lets us." Does it, though? Employees split into two speeds. Some shine, because their work automates beautifully and they want to automate it - suddenly the intern in the corner is doing the work of ten people. Others, by personality or by role, simply can't multiply their output. A new divide opens up. Different speeds, rising tension.
The result is a jittery market. A lot of founders live in permanent FOMO - "am I pushing my company hard enough? Am I missing something? Is my competitor already ahead?" I feel it too, and I run companies. So does almost everyone employing a few hundred or a few thousand people. The stress is real and it's climbing.
This is a guide for one specific reader. Not the founder starting fresh in the AI era - they have it easy, and I'll explain why. This is for you: 400 people on board, processes, departments, dependencies, managers. A team that's done the same job for five or seven years, sits comfortably in its zone, and is now - understandably - afraid for its role, its title, its job.
Starting fresh is cheating - so stop comparing yourself to it
Let me be honest about my own advantage, because it changes what my advice is worth to you.
Open Mercato is 7 people because we built it now, in the AI era. No legacy. No procedures slowing us down. At a company this small, the only "business" asset is culture - everything else, the process and the bureaucracy, hasn't been invented yet, and we're in no rush to invent it.
So we did two things. We hired people with a very high sense of agency (the drive to just make things happen, sprawczość), and we handed them AI. With no process to constrain them, anyone at Open Mercato can take a thing and build it. That combination produces output on the fly, often with zero process and zero procedure. If you're starting a company today, you're in a fantastic spot.
I'm fairly sure this holds at scale, too. Through the AI boom I've invested in around 40 startups. Some are already big - very big - because they grew fast, and they run exactly like this: tiny teams, high-agency people. I think a company can operate this way at 500, even 1,000 people. Being a Business Angel turned out to be an accidental MBA - I watched these companies up close for years, and now that I'm building my own again, I get that top-tier knowledge for free, no tuition.
But here's the asterisk, and it's a big one. Those are companies that grew to 1,000. Not companies that had 1,000 people and got transformed. That distinction is the whole ballgame. If you're transforming, you don't start from zero - you start from inertia. So stop benchmarking yourself against the AI-native darlings. Your problem is different, and genuinely harder.
Why bigger companies struggle - and it isn't about talent
In a big company, more people means - by simple math - not everyone is an A-player. That's not a judgment. It's impossible. You need the entrepreneurial ones and the detail-oriented ones. You need everybody. I'm not ranking human worth here.
But AI changes who can ride the wave. People who are more passive by nature - "tell me what to do, give me the procedure, I'll follow it" - won't tap the full spectrum of what these tools offer. Not because they're bad. Because they put a yoke on themselves: what am I allowed to do? They won't experiment fast with the tools sitting right in front of them.
So you get the painful ratio. 400 people, and maybe 10 of them jumped on the AI wave on their own. You want more to jump. And now the real question: do you want them to jump, or are you going to shove them in? And if you shove - will they swim, or drown?
That's the honest tension no framework deck will show you. Hold onto it. We'll come back to it.
You can't out-evangelize fear
There are three ways to convert an organization to AI, and they map onto the history of Christianity.
The Gospel. As CEO you walk the halls and preach: this is the moment, this is huge, let's step into the light side of the Force. Pure evangelism.
The baptism of Europe. A few centuries later, the pragmatic conversion happened- because it simply paid off. The message becomes transactional: do AI because you'll earn from it - bonuses, perks, promotions, you keep your job.
Fire and sword. The medieval method. Force. No volunteering.
Here's the line that hit me: the companies that succeed run all three at once.
We evangelize. We try to convert, we talk up the opportunity, we vibe-code (non-technical people describing to an AI what they want and getting it back). We made Claude a key "hire," and now we've got 30 internal vibe-coded projects - new products, internal tools. And I have this nagging feeling evangelism isn't enough. We're sitting there wondering whether the other two methods are for us or not.
Because the thing blocking people isn't a lack of belief. It's fear about their livelihood. And you cannot argue someone out of that with a better slide. The logical case for AI - "look how much more efficient this is" - lands fine. The emotional case is the wall: if I optimize myself, do I optimize myself out of a job?
I've lived a version of this before. When we sold my two previous companies, we learned you have to start communicating to people long before the event - the fear shows up the moment an investor walks in, and silence makes it worse. Same dynamic here. Address the emotion directly, early, or it runs the room.
The pirate ship vs. the tanker
When companies come to Open Mercato to adopt AI coding at scale, the story almost always starts the same way: the initiative comes from the board, driven by a positive FOMO. A CEO who's lightly energized - "are we missing something here, folks? Let's check." That energy is a great driver. Keep it.
From there I see two patterns.
Pattern one - the pirate ship. The CEO says, plainly, "I can't lift and change the big IT department, so we'll build a tiny one next to it." Run the numbers and that little pirate boat is maybe 10-20% the size of the main tanker. It often reports straight to the CEO. It's staffed with commandos who wanted to be there - recruited from inside and outside specifically for this one mission: we'll deploy AI in this giant bank / telco / you-name-it. It's exciting for them. They ship fast, because everyone on board chose to be on board.
And they're in open opposition to classic IT. Openly: "We're pirates, we do it differently, and maybe we'll convince the rest." The old IT becomes the source of truth - there's a SAP somewhere, you connect to it, whatever lives there is the truth, and you build your shiny new apps on top. This model works. Honestly, it works pretty well.
Pattern two - transform IT itself. Here the CEO, usually alongside the head of IT, tries to convert the existing department. And there's a brutal trap. The people doing the work know it better than their bosses do. So they can construct a thousand precise, well-argued reasons why it won't work. A massive wave of "this fails because…, and that fails because…." These clients call us and say, "My IT team won't buy it - can you explain it to them?"
That's an ugly spot to walk into. You've got the person who pays and badly wants it, and the people who have to use it and don't. Close to unwinnable. If you're early and unsure which road to take, start with the pirate ship. It de-risks the whole thing and buys you proof before you ever touch the tanker.
The technique that actually melts resistance
So how do we walk into that hostile room and not get torn apart? We do the opposite of what they expect.
We don't sell the big transformation. We say: let's not talk about changing all of IT. That's too much, it might fail. In fact - let's list everything that could go wrong. Out loud.
This builds trust faster than anything. We name the real risks first. The team goes, "Yes - and that won't work either, and that, and that." And then the magic line: "Finally, someone reasonable." There's so much bullshit on the market that simply being honest about the dangers makes you the credible person in the room. We even work slightly against our own interest - we're supposed to just sell a license, but we tell the board to slow down, because we know that if the project blows up, they won't renew. It's a long-term relationship, not a one-off license. The board hears that and trusts us more than their own people, because we're visibly more objective.
Then the concrete move. It's not genius. It just works.
We build a bypass. Today your people handle some task - say, customer tickets. We stand up a system that quietly routes 10% of that work to AI agents in parallel. Nothing goes live. Every ticket gets solved twice: once by a human, once by AI. The results are set side by side, and the human picks green or red - good answer, bad answer. Call it mirroring.
Run it for a month. Collect data. Then walk into the board with it. If the AI is genuinely useful in 30% of cases, you now have real numbers for that conversation - not vibes. And you'll discover something sharper: which subset of cases AI handles well. Maybe there's a cluster of tickets that are easy to resolve. You dig there, and you leave the rest alone.
Look at what just happened structurally. You moved the entire room from "who's right" to "let's see." It's a pilot. Nothing reaches production. Nothing breaks. And slowly, case by case, people stop believing AI is just for the crazies. That's the conversion evangelism couldn't buy.
One more reason IT is the sharpest example of all this: nobody saw it coming. A few years ago everyone in IT said, "We're the safe ones." Turns out their work is getting automated first. So there's a lot of genuine anger and emotion in that room. Respect it. Don't paper over it with a demo.
How to recruit for this
The people building Open Mercato did not come through a classic funnel. They showed up around the project.
That's partly a choice my brother and co-founder Piotr and I have in our DNA: when we build something, we tell people about it loudly, and that pulls in others who share the mission. We run open hackathons - one had around 70 people who came to co-build Open Mercato, because it's open source. You watch how someone works for two days, and the agency is undeniable. A person who, on paper, designs UI walks over and says, "I'll fix this interface - but completely differently. I'll write an agent to do it, because I'll learn something along the way." On paper they do X. In front of you, over two days, they show they can do Y and Z too, with joy, zero friction.
I see the same thing in The Five, the community I'm building around my book. People don't get recruited so much as they arrive: "Great mission, I was here, I did that, look at this." One person, instead of the CV everyone else sent, built us a working app. Different. More engaged. We invited her in and hired her the same day.
Here's the lesson, and it's counterintuitive. When AI hands out this much agency, when teams can be small and process-light, culture and mission matter more, not less. Your hiring funnel was built for a different era. If you force the best AI-native people back through it, you lose them - they self-select out, because they had ten other doors open. The signal you're hiring for isn't a polished résumé. It's the drive to build, and the proof they already did, ideally on your problem.
The structure: contracts, not org charts
If teams are this small and this capable, who manages them? Where are the managers?
What I see - in my own companies and in the 40 I've backed - is a shift toward internal contracts. A small team has an initiative. They treat the company as a platform that lets them do the thing they want to do. And the direction of the pitch flips. People come to us and say, "Great idea, it'll fit Open Mercato perfectly" - and they pitch it like we're an investor. We say, "You're right, that's strong. What do you need to build it?" They say, "One more person." Fine - go find them, or maybe you already have them.
Notice the reversal. In a classic big org, we tell you what to do and sell you the vision so you feel fired up. Here, you sell us. Then we strike a kind of contract: we give you time and resources for six months, and we agree on what we want at the end - revenue, lower cost, higher customer satisfaction. If it lands, great, we continue or start the next thing.
The rhythm is weekly, but it's not an interrogation. The team self-reports against the one core goal they own. The weekly cadence exists so we can support each other and unblock things - because plenty comes up along the way - not to grill anyone on where they are against target. The trust level is high because the goal is clear and the people are A-players who pitched their way in.
Big asterisk again: this works cleanly because The Five and Open Mercato are fresh. We don't carry the problem every larger company carries. So treat the contract model as the destination you're steering toward, not something you can paste onto 400 people on Monday.
What to do Monday
Don't transform your whole company. That's the mistake that produces a thousand reasons to fail.
Pick one team. Pick one repetitive task. Build a bypass that touches 10% of the work with nothing going live. Give it a month and one question: in what share of cases does AI actually win? Then walk into the board with numbers, not vibes - and ideally with the subset of cases where AI clearly wins, so your next step is obvious.
And stop relying on evangelism alone. Run all three methods at once. Preach it - share the wins loudly. Reward it - make it pay to adopt. And yes, eventually, mandate it. The companies that win don't pick one; they do all three together.
Then sit with the honest question, the one I can't answer for you. Of your people, how many jumped on the AI wave on their own? And for the rest - are you helping them jump, or shoving them in to drown? Get that answer right, and most of the transformation takes care of itself.
Tomek Karwatka builds Open Mercato, an AI-first ERP/CRM framework for teams adopting AI coding at scale. This piece draws on a conversation with Tomek Jabłoński (co-founder, Tutlo) on building teams in the age of AI.