AI Gave Us Lemons. We Picked Limoncello
Here's the thing. These days it's become genuinely hard to maintain professional motivation in software engineering. The level of bullshit is off the charts, everyone lies from companies to ChatGPT. AI rewrote this, AI replaced those people, AI this AI that blah blah blah. Today you're a lead engineer, people listen to you, tomorrow you're on the street under a bridge. Why? Well, random picked you. So what's the point of continuing to do engineering? Maybe I should not have listened to my mom and become a surgeon after all, but we are where we are.
AI really amplified this problem. Before you could at least stand tall on your skills, now that everyone is convinced three Claude Code agents can write any code, it's become very hard to prove you're something more than a useless bunch of cells, a layer between a manager and the real deal, the LLM. On one hand sure, we can separate work and life, you know, touch grass, do pilates, drink water. On the other hand the professional part of us is a big sore spot. Who am I? Another engineer who'll get tossed in the garbage? A cog in the machine? Even if you understand this is a transitional period between one stage of the industry and another, it's all incredibly demotivating. You lose all will to try.
Even before LLMs I noticed that it's very important to have some unkillable part of yourself that lying employers or a bad market can't take away from you, something that represents your professional self, a little piece of me in something. You know that dream about a cool pet project, or a blog, or teaching on the side? Why does that dream exist? Because a person needs some kind of foundation so that the next layoff or the next Super Cool Amazing Technology doesn't rip them out of the ground roots and all. Yeah, you can trample my garden and break my roof, but I have a basement full of canned food, I will actually survive. Yes, I'm a big fan of post-apocalyptic fiction.
And I have to say, despite everything, AI somehow improved access to building that basement. With all the quick solutions, cloud stuff and various deployment options, even a person who isn't super familiar with every stage of the process from idea to deployment can do it outside of work hours, build their own basement with canned food. On top of that it's an important part of professional fulfillment, setting yourself challenges like that, doing things you never get to try at work, seeing the full development cycle, plus trying to do your own marketing and sales. It's very sobering. It's hard to sell you bullshit and hard to sell yourself short after something like that.
In this article I want to share how we, a data engineer (me) and a full stack engineer (my brother, because who else would agree to this), built a cool (but of course completely unprofitable) project over three months, from idea to an actually running website with a product and what came out of it for us.
How It Started
Every developer at some point has that conversation with themselves where they go "I should really build something of my own" and then immediately open Netflix instead. We were no different, except this time there was a challenge involved and for some reason challenges work on engineers the way laser pointers work on cats.
We came across a challenge from Bright Data and n8n on Dev.to. If you're not familiar with Bright Data, the short version is: they give you access to the kind of data that big corporations created, collected, profited from and then hid behind walls from the very people who generated it in the first place. I'm a big fan of making things accessible and an even bigger fan of hating monopolies, so when I saw this my immediate reaction was "cool, let's do something with that."
The idea behind our Wykra is pretty simple to explain to a normal human being. Say you're a small brand or a marketer and you need to find creators on Instagram or TikTok who actually match what you're looking for. Like, vegan food bloggers in Portugal with 10 to 50 thousand followers. Right now your options are either scrolling for three hours or paying a platform that charges you like it's solving world hunger. We thought, okay, we can probably build something that does this, how hard can it be, famous last words obviously.
We submitted our challenge entry and then something unexpected happened: people actually liked it. Turns out I'm pretty good at selling an idea. The response was big enough that we looked at each other and said okay, this deserves more than a one-off post. That's how the whole build in public thing started. Because apparently the logical next step after "people liked our thing" is "let's commit to writing about it every week while working full time jobs, what could go wrong."
What It Actually Looks Like From the Inside
Here's something that happens when you build a side project as an engineer with a full time job: you suddenly discover that engineering is maybe 30% of the actual work.
We became our own product managers. What should this thing actually do? Who is it for? What's the MVP? These sound like obvious questions until you're the ones who have to answer them and there's no product person to argue with. We became our own QA. Which means we broke things, found the bugs, got angry at whoever wrote this code, remembered it was us and fixed them. We became our own designers. Please take a moment to appreciate the purple color #422e63 because I picked it and I'm unreasonably proud of it. We became our own marketers and let me tell you, this is where the real education happens. You can build the most elegant system in the world and then sit there watching your beautiful product get zero clicks because you wrote the landing page copy like engineers. Or, and this is my favorite example, you can launch your influencer discovery tool with GitHub login as the only authentication option. Because nothing says "welcome, marketing professionals" like asking them to log in with a developer account they don't have.
We also became our own researchers. Understanding the influencer marketing space, figuring out what people actually need versus what we assumed they need, reading about how competitors do things, learning that most of them are also kind of winging it. Very comforting honestly.
And then there's the build in public part. Week one: you write a post, it gets decent traction, people are interested, you hit top of the week on Dev.to, you feel like a startup founder giving a TED talk. Week two: still going strong, numbers look good, your motivation could power a small city. Week five: your cat is sick, you're tired, work was hell this week and you need to write something coherent about your progress but you haven't made any progress because life happened. Week seven: you skip an update and feel guilty about it like you missed a deadline at work except nobody is paying you for this. We kept a build in public series going for about eleven posts. The first two did well, after that the readership slowly settled into what I can only describe as a small dedicated group of people who clearly knew what they signed up for.
Somewhere around month two we almost quit. When your team is two people and one of them is your brother, every disagreement about architecture or priorities or "should we even keep doing this" carries about ten times more weight than it would with a coworker you can forget about after standup. There was a week where we barely spoke about the project and I genuinely thought that was it, we're done, this was a cute experiment and now it's over. The thing that pulled us back was embarrassingly simple: we'd already told people we were building this. The posts were out there, people were reading them - quitting quietly wasn't really an option anymore, and quitting publicly felt worse than just fixing whatever was broken and moving on.
The honest truth about build in public is that it requires either a wealthy uncle funding your free time or a very comfortable stock plan from a big tech company. For the rest of us it's a constant negotiation between wanting to share and wanting to sleep. But even with all that the discipline of having to explain what you're doing every week forces you to actually think about what you're doing. And that part is genuinely valuable.
The Practical Part
At some point you stop being a fun side project and start being a small business that spends actual money. You need a Google Workspace because you want to look like a real company when you send emails. You need API access to actually scrape social media platforms, which costs money because nobody lets you do that for free and for good reason. You need LLM calls for analysis, which means paying for models through something like OpenRouter. You need infrastructure to run all of this, which means a hosting platform. You need monitoring because things will break at 3am and finding out from an angry user is significantly worse than finding out from an alert on your phone.
I'm not going to drop exact numbers here, but I will say this: if you're smart about how you use LLMs, pick the right model for the right task instead of throwing GPT PRO at everything, cache aggressively and keep your architecture simple, the total cost of running something like this is surprisingly manageable. We're talking "a couple of nice dinners" territory, not "second mortgage" territory. The time cost is harder to quantify. Three months of evenings and weekends, some more intense than others. The average is probably somewhere around ten to fifteen hours a week if you spread it out, which sounds fine until you remember that those hours come after your actual job.
But say you're reading this and thinking okay, I actually want to try. The problem is that the gap between "I want to build something" and actually building it is enormous and most people live in that gap forever, thinking about it in the shower and then doing absolutely nothing about it. So let me walk you through how we got past it, because looking back it's less mysterious than it felt at the time.
Coming up with an idea is the part everyone overthinks. People sit around waiting for some brilliant original concept that will disrupt an industry and that waiting period conveniently never ends. Look at what annoys you, look at what annoys people around you, look at challenges and hackathons happening online. We literally found ours because Bright Data posted a challenge and we went "sure, why not." Your idea doesn't need to be revolutionary, it needs to be something you can explain in two sentences and something you care enough about to still want to work on at 10pm after eight hours of your actual job.
Starting is the part that feels impossible and is actually the simplest thing in the world. And here, ironically enough given everything I said in the introduction about AI making our lives miserable, is where LLMs actually become incredibly useful, but probably not in the way you think. Forget about asking them to write your code. Instead, open a chat and say "I want to build a calorie calculator but I have no idea where to start, be my coach." Tell it what you know, what you don't know, what scares you about the process and let it walk you through it step by step. Ask it to break down the project into pieces small enough that each one feels doable on a Tuesday evening. Ask it what to build first and what to ignore for now. The same technology that's causing all this professional existential dread turns out to be the best free project coach you've ever had and the universe clearly has a sense of humor about these things.
Keeping going is where it gets genuinely hard because the initial excitement wears off somewhere around week three and suddenly you're staring at your codebase on a Friday evening thinking about all the other things you could be doing with your life. If public accountability isn't your style, find a friend, a Discord server, a coworker, literally anyone who will periodically ask "so how's that project going" and make you feel just uncomfortable enough to continue.
Spending money is the moment where your brain starts negotiating with you. You need a domain, hosting, API access, some LLM credits and the voice in your head goes "wait, we're spending real money on something that might never earn a single dollar back, is this wise?" Push through that voice. The amounts involved are genuinely small if you're smart about it. Platforms like Render or Railway or Fly.io will host your thing for the price of two coffees a month. OpenRouter gives you access to LLMs without requiring a second mortgage. Cloudflare will sell you a domain for less than lunch. And honestly, a huge thank you to the companies that offer free tiers and there are some open-source models, because for people where money is the thing that blocks them from even starting, this matters more than those companies probably realize. I'm personally a big fan of Neon for databases and Streamlit for quick prototyping, both of which let you get surprisingly far without paying anything at all. You have absolutely spent more money on things that brought you less satisfaction, I can almost guarantee it.
Deploying used to be its own special circle of hell but in 2026 you can push your code to GitHub and have a live website with a real URL in minutes. Vercel, Railway, Render, pick whichever one you like, connect your repository, hit deploy and watch it happen. If you've never done this before it genuinely feels like magic the first time and the important thing is to do it early and do it ugly, because a running ugly thing that real humans can actually visit is infinitely more real than a beautiful polished masterpiece sitting on your localhost that nobody will ever see.
And that's it really. Idea, start, keep going, spend a little money, deploy. It sounds like a lot when you list it out but each individual step is something you can figure out in an evening and before you know it you have a real thing running on a real URL that you can show to real people. The whole process is less about talent or genius and more about stubbornness and refusing to stop when your brain is begging you to go do something easier. Which, if you think about it, is basically what engineering has always been.
Limoncello
So after three months of building, writing, debugging, designing, marketing, spending money and occasionally questioning our sanity, what did we actually get? A running product that nobody uses and the knowledge that we can do the whole thing from start to finish. Some of it well, some of it in a way we should probably never examine too closely, but all of it done.
We collected eight stars on our GitHub repository. Eight. I'm going to put that on my resume.
We also discovered that this resonates with people way more than we expected. The frustration, the desire to own something professionally, the need for that basement with cans. A lot of people feel this and a lot of them want to do something about it but get stuck in that gap between wanting and doing. The fact that we went from "this would be cool" to a real running website was apparently inspiring to some folks, which is both flattering and a little sad because it really shouldn't be that rare.
As for what's next, right now I'm packing a suitcase to fly to Japan for two weeks and air out my brain and after that we'll see. Our Instagram search is still not great, we still don't search other social networks or Google Maps, I'm still unhappy with the level of analytics we provide and I still need to learn the subtle art of attracting users without attracting the attention of a psychiatrist. We also have the option of launching on Product Hunt and Y Combinator has a round in April, so who knows.
But here's the part that matters more than any roadmap. The reality that made everything worse also handed us the tools to build something of our own. The same AI that threatens to replace us helps us prototype faster. The same cloud infrastructure that big companies use to run their empires is available to two people working after dinner. The same internet that's full of doom and gloom about engineering careers is also full of people who want to see you build something and will cheer you on while you do it.
You are going to get squeezed by this industry, that part is probably unavoidable. You're going to get handed lemons. The standard advice is to make lemonade, smile, be grateful, pivot, adapt. We chose limoncello instead, because it takes longer, it's more work, nobody asked for it and in the end you have something with a bit more kick to it. It will cost you your evenings, some of your money and a lot of stubbornness. The result might be completely unprofitable. But you'll end up with a basement full of cans and when the next storm comes, and it will, you'll know you can survive it.
If you want to dig around in our basement: github.com/wykra-io/wykra-api
And if you want to see what else I write about when I'm not making limoncello: datobra.com