The unvarnished truth about building a category-defining AI company

It is 4 am in Lviv, West Ukraine. I am on a green foldable sleeping mat in the bunker in our apartment complex. TripAdvisor: "Five stars. Would hide from missiles here again."

We are hiding from Russian ballistic missiles. The air raid siren went off 20 minutes ago. Simba, my Shiba Inu, has decided this is a personal attack on her sleep. She's judgementally glaring at me like I owe her money. I can almost hear her think: 'I need my beauty sleep, you piece of ****.'

In five hours, I have fundraising meetings with some angels. And a meeting with the Partner of a world-famous VC firm. My wife and I are on Telegram tracking the missiles and Shahed drones. Trying to figure out when we go back upstairs and sleep.

I am mentally visualising my pitch:

So, about our Neurosymbolic AI...' The Partner could not make it. So instead I am explaining the future of enterprise AI to a 24-year-old associate who graduated last spring. He is quietly Googling 'neurosymbolic' while nodding. But he's British, so he says 'fascinating.'

Fast forward to 2030. Pebbles Ai goes IPO. The Partner at a dinner party reminiscing with someone that they 'almost invested early' in a category-defining startup. The associate, now a barista, occasionally Googles 'neurosymbolic' between orders. Still has no idea what it means.

Anyway, the actual pitch went well. Nothing says product-market fit like a founder who has not slept since Tuesday. But the occasional missile attack is easy-peasy compared with building a category-defining AI company.

That is a whole new level. To compare apples to apples, it is like sitting in the same bunker. But with 20 devoted Hare Krishnas singing the same $%^&*% Hare Krishna song for 5 hours straight.

Very different than running a strategy consultancy in London. Back then, I was making 7-figures in gross margin in my first year with excellent work-life balance. Hell, I even had a pilates teacher. How Soho posh of me.

But darn my pattern recognition brain. I saw what was coming on the horizon. It was not only the market opportunity I noticed. I also saw something else, which terrified me. My Scrooge McDuck in me wanted to capitalise on it. Clearly a perfectly fine healthy reaction.

I saw a future where AI systems that cannot reason replace the professionals who can. Marketing and sales hollowed out by automation that produces volume without results. Millions of white-collar jobs evaporating. Not because the work stopped mattering. Because everyone believed that mechanical tools and general-purpose AI were better than human reasoning in complex verticals.

I knew that the answer was a Neurosymbolic AI built into an operating system that anyone could use. This was way back in 2021. And just the right time to have that idea.

So, I spent 18 months writing the scientific blueprint of our AI today. After that, I thought I would just build the product, open a company, gather a team. And raise £20m. Then, I remembered I was in Europe.

And in Europe, people see problems, not possibilities. ‘Here is a baby unicorn.’ ‘Hmm… not sure about the size of that horn.’

Also, in my second venture, I learned you do not run one type of company. You run a series of completely different companies at different time stages that happen to share a name.

Stage one: just you, an idea, some legal scaffolding, and a prayer. Stage two: different team, different tech stack, different product, different everything. Stage 3: you're basically a car dealer salesperson with a pink tie and whitened teeth from Turkey.

You are basically the main character in a game who never gets to chill. Each stage unlocks new boss fights. New complexity. New costs. New catastrophes. While the arena keeps getting bigger. You start missing the early days when it was only about the scientific research.

Now you have to worry about the full product lifecycle, product-led growth (PLG) journey, marketing strategy, sales processes, legal, finance, and accountancy. Not to mention, the always present existential dread. That last one is not a department. But it takes up more of your calendar than any of the others.

You have just enough money to assemble something resembling a team. Not enough to hire a CFO. Not enough to hire a COO. So guess who becomes both? The Founder.

The person who was supposed to be doing product things is now doing every other job in the company.

If you are lucky, some amazing early-stage angels step in. They help with the operational, financial, and legal work. If you are not lucky, you learn it all yourself. At 2am. From YouTube. I have done both. YouTube is surprisingly good at copyright law. Less good at helping you sleep afterwards.

And on top of all of that? You deal with people. Their feelings. Their crises. Their needs. You become the therapist, the cheerleader, the strategist, and the asshole who fires them because they cannot do even one thing right. (#youonlyhadonefuckingjob).

I once gave a motivational speech at 10am. Restructured a research paper at noon. Fired someone at 3pm. And made Dutch pancakes for the team at 5pm. Because morale needed sugar more than my speeches.

That is the baseline. That is the job. Before you add the small matter of building a system that has never been built before. And we all know how Neurosymbolic AI is easy to build. Especially when it requires feeding from different fields of science. 

From persuasion sciences and propaganda to behavioural economics. Easiest thing in the world. (Sarcasm)

Then, you learn that it is not actually the long hours and the deep work that make you tired. It is the constant context switching. And the realisation that the difference between a £10m company, a £100m, and a £1bn? The number of sequential right chess moves you make.

It is like growing up. You are a kid. You do stupid things all the time. You are teething as you grow. Bones ache as they get bigger. Nobody looks at a toddler and thinks ooh, he needs a bit of empathy. You just get a slap when you make the wrong move.

And boy did I need slaps.

Growing Pebbles Ai is exactly the same. I am back to being a toddler. Except the teething is 50+ engineering sprints and shipments to get to what resembles a product that is foundationally different.

And the growing pains are not just technical. They are compliance frameworks, GDPR, data protection across jurisdictions, and dealing with difficult customers with humility.

Not to mention your first talent pool, which is comprised of a bunch of raccoons and hyenas. Don't expect anyone normal to apply for your vacancies.

Top talent, the creme de la creme, only comes if there is something worth coming to. It is a rite of passage every Founder has to go through. It is painful. And in our first year we had little to show for.

And then of course, the war on top of it. Power cuts. The buzzing.

Nothing reminds you more of your mortality like Shaheds. Except building an AI startup. That takes the cake. Ever sold a Beta to a customer? I actually prefer the Shaheds.

But I knew that I would find my people eventually. Ukraine has real generational STEM talent. The best engineers in the world outside of Silicon Valley are here.

A culture that takes pride in technology. Values that run deeper than a mission statement on a wall. And the quiet ritual of sharing good borscht after a long sprint. And there is something romantic about being in it together.

Fast forward to now. I found my team. An all-star squad of A-players with IQ pouring out of their ears. Lion-hearted courage. Relentless intellectual curiosity. And the kind of grit that investors dream off.

Here is one story that perfectly illustrates my team. A Shahed drone detonates 400 metres from the Pebbles Ai office. First time since the start of the war that one hits that close. Team goes to the bunker. Nobody logs off. They are just annoyed they left their fancy keyboards and curved monitors upstairs.

So they do what any sensible team would do. They close some Jira tickets on their laptops.

One of them submits a pull request from the bunker. Cleaner than anything I have ever seen written. Even in my old Google office with central heating.

That is a Tuesday for this team. A normal Tuesday. That is not culture you manufacture in a WeWork over a flat beer and hipster sneakers. That is culture literally forged under fire.

Personally, I could return any time. London. The Netherlands. But I do not go back. Because leaving would mean leaving my team. And my team is everything to me.

That is not noble. It is actually selfish. I need them more than they need me. Having them around makes this journey worthwhile. Do not tell them I said that.

But the loneliness never leaves. Nobody wants to talk about this part. It is not very sexy or “manly.” But very much a daily reality.

It is profoundly, structurally, architecturally lonely. I use three adjectives to show you I have a rich vocabulary.

Your team cannot know everything. Existential threats appear on the horizon. Sometimes twice in an afternoon. If you told your team every time, half of them would walk out before you finished the sentence. Not because they are weak. Because they are human. Constant ambiguity is not a natural human habitat.

So you protect them. You absorb it. You sit with it at 11pm. Staring at a spreadsheet that says you have three months of runway left.

You Google ‘how to extend runway without selling a kidney.’ You close the tab because the results are not as metaphorical as you hoped.

You go to bed. You do not sleep. You get up. Start calling investors. And give your team the headspace to continue building with focus so we get to a product that customer can actually benefit from.

Besides, your investors backed a winner. They do not want a tearful voice note at midnight. They want the stoic founder. The one who keeps buggering on.

You send the yearly investor update. “Strong progress across all workstreams.” Translation: I have aged six years in the last 12 months. The direction is forward. First revenue. Early PMF. Thank god. The hard work and the right moves are slwoly paying off.

Though the compass is still broken and the map is on fire, at least the vibes are cautiously optimistic. And we are in it together. One Leonidas is now a 10-team strong Spartan squad.

Two years ago it was Leonidas and a bunch of raccoons. I call this progress.

But the loneliness still remains. Your family loves you. They want to understand. They genuinely cannot. Not their fault. It is a different world entirely.

My wife comes closest. She is hiding from the same missiles. She sees the 4am face working on our financial model. But even she cannot feel the specific weight of knowing that every person on your payroll trusted you with their career. And their livelihood.

You are one bad quarter away from betraying that trust. That weight has a texture. It sits in your chest. It does not leave.

Only other, older founders get it. The ones who look at you, nod slowly, and do not need you to explain why you have that specific look in your eyes.

Only other, older founders get it. The ones who look at you, nod slowly, and do not need you to explain why you have that specific look in your eyes.

Those people are rare, busy, and deeply precious. If you are lucky, you have one.

And then there are the people who watched Pumped and The Playlist and now think they understand startups. They saw the montage. The highs. The dramatic lows. The triumphant ending scored by an indie band.

What they did not see was the 1,000 days in between where nothing cinematic happened. Just a founder and a team working 15-hour days, 6 days a week. Grinding through complexity that would make a Hollywood screenwriter quit. Too boring to film.

'Why is it not scaling yet?' Because neurosymbolic AI is not a WordPress plugin. 'When will the product be ready?' When it is ready. 'Can you move faster?' We are already moving at a speed that would hospitalise most people.

Go too early without a moat. You win at first. You lose later. Go too late. The market moved on without you. Go without the right team. Customers pour in and you collapse under your own weight. Wait too long to launch. Faster competition takes the brand name.

The trick is finding the right moment. Not too early. Not too late. With the right people. With the right foundation. With enough scar tissue to know the difference.

We found ours. And it's now!

That is why often the best investors are entrepreneurs. They lived it and they know what the sexy montage skips.

They have empathy and sympathy that fuels the heart, and strengthens the spine.

But even if you have amazing investors, which we do, as a Founder you still constantly fluctuate between highs and lows. If someone tells you they do not experience this, they are either lying or they are not pushing hard enough. Or they are a psychopath. In which case they are better off working in investment banking.

The loneliness is one thing. The technical isolation is another. We did not pick the easy path. It has never been done before. And when I say that, I do not mean “nobody has done it quite this way.” I mean nobody has done it. At all. Ever.

It is like you never had a dog, so you go and buy the most difficult breed: a Shiba Inu. ‘Hey, let us start a Neurosymbolic AI company. A SaaS company is too easy’.

If you didn't know, most AI companies take a large language model. Wrap a UI around it. Add a logo. Call it “AI-powered.” That is not intelligence. That is a skin.

A very expensive, very risky skin. The kind that raises Series A rounds and produces output that reads like it was written by a committee of chimpanzees high on Adderall.

They don't often do so well in the long term. Especially after VCs find out there is no tech depth. No defensible IP. Nor a plan to create one.

This requires deep subject matter expertise. A handful of people have it. And they often do not quit their six or seven figure jobs to start this entrepreneurial journey of hell. They are good where they are.

All of these go-to-market sciences, expertise, and wisdom are hoarded. Inside big tech companies like Google and top-tier strategy consultancies like McKinsey.

They have the best technology that acts as the conduit. The best scientific knowledge that fuels the engine. And the best talent who are the operators. Not to mention, a legal team the size of a small country to make sure nobody else gets any of it.

Until now. David did not beat Goliath by getting a discount on Goliath’s sword. He brought a completely different weapon. That is what we are. A completely different weapon. A new paradigm. A GTMOS™ built on neurosymbolic cores.

Building a category-defining neurosymbolic AI company is not for everyone. It requires insane levels of IQ. The kind that makes me the dumbest person in the room. And I am fine with that. I am spectacularly fine with that.

In fact, assembling this core team is one of the most important professional achievements of my life. Protecting them is my daily obsession. Making sure their belief was justified is the thing that gets me out of bed every morning.

So why do we all do this?

This increased oligopolisation has real consequences on real lives.

Microsoft bought LinkedIn, GitHub, and Activision. If they could acquire oxygen, they would charge you a monthly subscription to breathe. With a 20% annual discount.

HubSpot will give you a free CRM. Then charge you for the privilege of making it actually do something. The first hit is always free. Ask any Dutch drug dealer. Hypothetically.

Salesforce’s mascot is a friendly cloud character. How adorable. Nothing says ‘we will bleed your budget dry’ like a cartoon bear in a Hawaiian shirt.

The average enterprise SaaS contract has more lock-in clauses than a prenuptial agreement. And somehow still less romantic.

This all affects lives. Take a brilliant founder in New York who watches his life savings disappear. He cannot get the word out because he has almost no budget for GTM despite having an awesome product.

Or the case of a marketing agency that feeds 20 families in London, who now lose their income in one month. Not because their company failed. Because the market turned and they did not have the tools to respond.

Or a Ukrainian IT company that loses 80% of its team. Not because of anything they did wrong. Because the tools to hedge were unaffordable, and they couldn't do a roadshow due to travel restrictions.

Three different cities. Three different stories. The same structural failure. All real.

Meanwhile, a Partner at McKinsey just invoiced someone £400 an hour to recommend ‘a more customer-centric approach.’ Groundbreaking stuff.

Salesforce costs more per year than some of these companies make in revenue. And you need 10+ freaking tools to have some decent go-to-market motion as a business that is not in the Fortune 500.

This is not a market gap. It is a democratic problem. A healthy economy needs a thriving middle. A strong class of SMEs and mid-market companies who employ the young, train the next generation, and prevent monopolies from hollowing out capitalism itself.

Pebbles Ai exists to change that. One platform replacing 10+ tools. We are the GTM firepower companies need to win back market share from the Goliaths.

So after all of that. We are live!  And we built something we are genuinely proud of.

Go try it out for free!

Written by Emin Can Turan, Founder and Lead Researcher at Pebbles Ai

‍