I am a software engineer currently doing my PhD in AI and I have been interested in entrepreneurship since I can remember and the idea of "making it" myself has been the main driving factor. But I am doing everything myself and frankly I am also a black sheep in my family, because I always wanted to do things differently. I do not come from an environment that encourages entrepreneurship (or investing for that matter…) so I needed to learn everything from scratch. I have done countless attempts of putting my website or app out there, just to see myself fail yet again.
At this point I think I became comfortable with failing and the thought of success stresses me (I know it is weird). Last year I needed to scrap a project that I have been working on for 2.5 years without any revenue, not a single sale.
I have learned from my mistakes and this time around I actually have validated market and work together with the target audience. These are still early stages of the current project though. I would just like to know if there is any secret ingredient to doing marketing, SEO or whatever else that make people succeed? Also how don't you overthink things and worry if you are a single developer that owns a website about all the things that can go wrong? How do you listen to feedback and how do you choose what to implement and what not?
I’m an AI Engineer who is great at building and terrible at business. How do I break the cycle of "Build, Fail, Repeat"?
byu/wojaczek28 inEntrepreneur
Posted by wojaczek28
4 Comments
Fellow engineer here who went through the exact same cycle. Few things that actually moved the needle for me:
1. The 2.5 year project with zero revenue is more common than you think. The mistake is usually building in isolation without a tight feedback loop. Sounds like you already learned this since you mention working with your target audience now, which is huge.
2. There is no secret ingredient for marketing. But there is a pattern that works: pick ONE channel (Reddit, Twitter, cold email, SEO, whatever), go deep on it for 90 days, and track what converts. Most engineers spread themselves thin across 5 channels and get zero traction on all of them.
3. On overthinking: I used to spend weeks agonizing over feature decisions. What helped was setting a hard rule of shipping something every week, even if it was small. The feedback you get from real users will tell you what to build next way faster than your internal debate ever will.
4. For the “what can go wrong” anxiety: make a list of the actual worst-case scenarios. Most of them are recoverable. Server goes down? You fix it. Someone finds a bug? You patch it. The real risk is spending another 2.5 years on something nobody wants, and you already have the antidote to that since you are validating this time.
The fact that you are comfortable with failure is actually a superpower most people here would kill for. Just make sure you are failing fast and learning, not failing slow and burning out.
I want to touch on a topic that may be a bit uncomfortable, but it’s something I’ve noticed a lot among strong technical founders. Often, the cycle of “build, fail, repeat” isn’t simply due to a lack of skills; it’s actually a matter of control.
When you’re building, everything feels predictable. Code behaves in a logical manner, and you can make tangible progress every day, which gives you a sense of competence. However, stepping into the realm of business can be jarring. It immerses you in an environment filled with emotional feedback that can be inconsistent and, at times, downright unfair. You might follow all the right steps and still face silence, which can be psychologically taxing for those who thrive in deterministic systems.
As a result, the mind tends to gravitate back toward building, it’s the last domain where you feel you have complete control. The founders who manage to break free from this loop don’t necessarily learn secret marketing tactics first. Instead, they start by changing what they measure.
Instead of asking, “Did this project succeed?” they shift their focus to, “Did I receive one irreversible signal from a real buyer this week?” This new perspective emphasizes metrics like actual payments, commitments of time or resources, or changes in behavior because of the product. It may seem like a small tweak, but it fundamentally alters decision-making. You begin to build not for elegance but toward those moments of commitment.
Ultimately, many projects don’t fail due to poor execution; they fail because they never prompted reality to respond strongly enough.
man the 2.5 years with zero revenue… i’ve been close to that and it’s brutal. the thing nobody tells you about being technical is that building feels productive even when it’s not. you can spend 8 hours coding a feature and feel accomplished, but if nobody asked for it you basically moved backwards. that’s the trap.
since the others already covered the marketing channel stuff, let me hit on your feedback question because that’s where i see most technical founders get stuck. here’s what actually worked for me: keep a simple spreadsheet of every feature request or piece of feedback. tag each one with who asked (paying user, free user, random person). after a month you’ll notice like 3-4 things that keep coming up from people who actually pay or would pay. build those. ignore everything else. sounds obvious but most devs just build whatever the loudest person in their inbox is asking for.
also for the overthinking… as an AI person you probably already know this but think of it like training a model. you need data to optimize. sitting there theorizing about what might work is like trying to tune hyperparameters without running any experiments. just ship something small, measure, iterate. your PhD brain wants the optimal solution before starting but in business you find it by running experiments on real users.
one more thing since you’re in AI specifically… the market rn is insanely noisy. every other person is launching an AI wrapper. the ones that actually survive are solving very specific painful problems for very specific people. “AI tool for X” is too broad. “AI tool that helps divorce lawyers draft discovery requests 3x faster” is a business. get that specific and you won’t need to worry about SEO tricks.
I was in the same spot. Built an AI document chatbot, spent months on it, zero sales. Product worked fine, just had no idea how to get it in front of anyone. Didn’t even know what an ICP was.
There’s an order of operations most technical founders get backwards (I did). Build first, figure out distribution later. Should be the other way around, or at least both at the same time.
What’s actually worked for me: hanging out in communities where my buyers are and being useful without selling. First paying customer came from an Indie Hackers post where I was just writing about how I figure out who to sell to. Wasn’t pitching.
Cold outreach, threads, directory submissions, none of that has gotten me a sale. Community stuff got me one in 2 weeks. Not saying that those other channels don’t work, but they haven’t for me yet.
You said you’re already working with your target audience so you’re ahead of where I was. Just keep showing up where they are and talk about the problem, not the product.
I ended up building a framework around this whole process after going through it, AI agents that do the research and outreach parts so I can keep building. Using it for all my products now.