Jakub Czakon - founder of Developer Markepear and former CMO of Neptune.ai (acquired by OpenAI)
Jakub Czakon (Kuba) on eight years marketing Neptune.ai to its acquisition by OpenAI: why he'd rather copy his users' own words than write cleverer copy, how being from Poland instead of the Valley shaped their whole strategy against Weights & Biases, and why the real marketing failure is silence, not backlash.
Jakub Czakon, known as Kuba, was until very recently the CMO of Neptune.ai, which was just acquired by OpenAI. He's also the writer behind Developer Markepear, a newsletter on DevTools marketing, and a strong enough chess player to have his own Wikipedia page.
We asked him to reflect on eight years of marketing Neptune, from choosing which market to chase to how they beat a much better-funded, Valley-based competitor by playing a completely different game — plus the writing habits and copywriting tricks he's picked up along the way.
The Q&A below is experimental — extracted from the episode transcript with AI. If you spot a mistake, let us know at jack@scalingdevtools.com.
Looking back at Neptune's journey to being acquired by OpenAI, what's the biggest lesson about product-market fit?
I'd just listened to your last two episodes before this one, both on product-market fit, and it crystallized something I sort of knew but hadn't fully internalized: if you market a product into a market that's hot, that wave carries you so much further than marketing into a cold one. That's not the same as being niche — I actually love niche markets, because you can serve people within them much better. Neptune was around for eight years, marketing to MLEs, data scientists, and AI researchers, and those personas shifted in importance over time. More recently we doubled down hard on foundation model labs training frontier models — a genuinely hot market, even if there weren't that many of those companies, because there are more verticals in that space than people expect. When a market is well-funded and the problems are clear and expensive, you get pulled into it, and if you pick well, sometimes by sheer luck, everything downstream — messaging, positioning, channels, campaigns — gets so much easier. It's easy to forget that and get sucked into your own vision for the product instead of noticing you're moving too slowly into something with real pull.
Investors push founders toward a big vision and a big TAM. How do you reconcile that with what should actually be on your homepage?
You need both at the same time, but they're different things for different audiences. Investors are right to push for a big market and a big vision — the market matters more than the product or the team, in my opinion. But your actual homepage visitors don't want a vision, they want the narrow thing you have right now. The bigger and more ambitious the positioning, the more it implies a big migration, a big platform, a big project — and early on, that's almost never good. I see founders fall into first-principles thinking here: what's the five-year solution, how do big DevTools talk about themselves, and then they put the biggest possible version of their vision on the website. I think that's a real mistake almost up until you're about to IPO, because you're missing out on people you could actually get, retain, and make happy today by talking in vague, hand-wavy terms instead. The vision belongs with investors, maybe on an About page — not as the whole website. Positioning should be based on your reality today; the more enterprise and more PLG you are, respectively, shifts how far into the future you can honestly reach, but it should still be grounded in what you actually have.
You've got a test involving a founder's h1 and subhead — what is it?
I ask founders to look at their own website header and subhead, then ask them, separately, what does your product actually do? Eight or nine times out of ten, their spoken answer is completely different from what's written on the site. When I ask why they don't just use that explanation, it's usually because they wanted something that resembles the big companies they admire — a flavor of Stripe or GitLab. But you shouldn't copy that, especially early — go look at Stripe's website on the Wayback Machine from 2014 or 2015, it's completely different from today, because brands change as they grow. When you use language that doesn't actually describe who you are, what you do, and how you're different, you put the entire burden of figuring that out onto the person visiting your site, and that doesn't help you.
You've talked about using your customers' own words instead of writing cleverer copy yourself. Where does that come from?
One trick: ask your existing users how they'd explain your product to a developer friend, and listen for patterns. Often the answer is completely different from — and better than — the words you were using yourself, and if enough different people land on the same explanation out of everything they could have said, that's a signal it belongs in your header. I do the same thing on Reddit and Hacker News: I go read threads in a category I'm working on and look for the memes, the language, the concepts people are already using, then adapt that to whatever I'm trying to communicate. There's an idea from a favorite copywriter of mine, Harry Dry, that a good copywriter doesn't really write, they point — your job is to point at an idea that's already out there rather than invent your own, more complete version of it that only a small fraction of people already understand. You might be more right, more complete, but it won't resonate, because it doesn't connect to what people already know. The best converting copy is almost always just copying what your own users are already saying, in their own words.
How did you figure out how to compete with Weights & Biases, who were better-funded and deeply embedded in the Valley?
Everybody's dealt a different set of cards, and you have two options: play the same game as your competition, or figure out what they're not doing and take that instead. Weights & Biases played their cards really well — Valley-based, well-funded, partnered early with OpenAI on the research side, deeply embedded in meetups and conferences. We were from Poland. If we'd tried to compete on Valley meetups with our seed round, we'd have lost that game outright. So we asked what they weren't doing, for whatever reason, and went hard at SEO instead — owning the blog, the whole MLOps tooling landscape, building authority in adjacent categories we could later connect back to us, deeply researched how-to articles. We crushed that. If your competition isn't doing something because they have other, better cards to play with their time, that's your opening — figuring out how you specifically can win, given your actual cards, matters as much as picking a hot market in the first place.
Looking back, what do you think you got wrong strategically?
The biggest one: it's almost always better to do less, but push it further until it saturates. We did a lot with SEO and the blog, but I think we diversified too early instead of really owning it completely — for example, YouTube search is genuinely great for SEO and we never built there. Once you find something that's working — getting good signups, good pipeline, happy activated users — you should be able to pour in more resources until you genuinely stop seeing marginal returns, rather than spreading into new channels. We could have made every single article ten times better instead of moving on once we were ranking well. As a former trader, I think about it like finding a trade you can exploit — you plow in as much as you can until the market becomes efficient and it stops being an opportunity. Most things you try as a startup won't work, which is expected, but when you do find the thing that works, focus and doubling down beats diversifying too soon.
You've said the real failure in marketing is silence, not backlash. What does that mean for how much risk you'd take?
In marketing, results are heavily skewed — the failure state is nobody hearing you at all, not people disliking something on Reddit or Hacker News, which usually isn't that bad. I think it's almost always better to aim for something that could be terrible or great rather than something safely average, because the average outcome online basically doesn't get seen — the big hits are what get you discovered. A concrete example: at a conference, we considered putting stickers down near pedestrian crossings near the venue, decided it was too risky because of city fines, and didn't do it. Another startup did something similar and clearly got a lot of attention for it. I later realized, during Comic-Con at the same venue, that big hotels were putting up huge unauthorized banners and just paying the city's daily fine as a cost of doing business, because the exposure was worth it. I wish I'd taken a bit more of that risk — a bit more aggressive, a bit more personal or snarky in some blog posts and ads — because you have a real chance of an outsized payout, and the downside is usually smaller than it feels in the room.
You used to hate events. What changed your mind?
I used to think events were a complete waste of money. I've completely flipped on that, especially with how much more people are craving real in-person interaction now. Good invite-only side events during big conferences worked really well for us — connecting the right people, doing something genuinely fun, and putting the brand out there in a way that isn't just a booth. It's still consistent with playing to your strengths, though — there are only a handful of events everyone in your space actually goes to, and instead of a generic booth, you can take that same budget and do something different there.
You mentioned front-loading information as one of your biggest writing lessons. What's the idea?
It comes from a book I recommend a lot, Writing Without Bullshit — the reverse pyramid idea, like you're on a battlefield and your communication line might get cut after one message, so you send the most important thing first, not "how's your wife doing." I see founders build up to their point instead of leading with it constantly, both in their own writing and when I'm working with them now. A good test: imagine reading your article out loud, word for word, to the actual people you're trying to reach, standing in a room with them at a small meetup. Almost every time, there's zero chance you'd actually talk to people that way — you'd sound ridiculous — and yet people write that way constantly. The fix is to front-load the core information first, at the top level, and then expand each point below it, almost like a TL;DR for every section as well as the whole piece — what does someone need to understand if they only read this section, and what do they need to understand if they only read the intro? Great speakers do the same thing: tell people up front what the few points are you're going to make, hint at the value in each, then go deeper — so people can follow along and don't lose the thread halfway through.