A crowded market is not a closed market. It is a list of incumbents who got comfortable. That is the whole reason I'm building another social media tool, and the only honest one I have.
I spent most of my career writing software that solved problems I didn't have, for people I would never meet. Sydium is the first thing I've built where I am the user. I don't have to imagine what's broken. I live it.
So when people ask why a solo developer in Romania would enter a category with fifty tools already in it, the answer isn't a clever wedge or a market gap on a slide. It's simpler and harder to argue with: I needed the thing, nobody else built it the way I needed it, and I can write code.
"Aren't there 50 of these already?"
Yes. And that is an argument for building, not against it.
The social media management market is worth $32 billion and growing at 24% a year. Fifty competitors in a market that size doesn't mean "stay out." It means people are paying real money, and the tools taking that money stopped being hungry years ago.
I researched all of them before I wrote a line. Buffer, Hootsuite, Later, Metricool, Sprout Social, Publer. The pattern was the same every time: the tool does about 70% of what a creator needs, then makes the last 30% painful, or it does everything and costs more than my rent.
Here is the specific thing I kept noticing. Buffer was revolutionary in 2013. Hootsuite was the answer when "social media manager" first became a real job. Both were architected before AI could write a usable sentence, before one person could realistically run six platforms alone. When AI finally arrived, most of them bolted a "generate caption" button onto a 2015 scheduling screen and called it innovation.
I wanted AI to be the foundation, not a button. Sydium learns how you actually write - your vocabulary, your rhythm, your odd emoji habits - then writes in that voice, schedules it, and publishes. You review once a week, or you let it run on autopilot. I wrote an honest breakdown of how that compares to Buffer and Hootsuite, including the places where Sydium is still worse.
The lesson that reshaped the whole product: distribution beats creation
I built a good tool for making and scheduling content. Then my own data told me I had been solving the wrong half of the problem.
At its peak my Twitter account was pulling around 332,000 weekly impressions, with zero ad spend. Almost none of it came from the posts I scheduled. It came from replies - showing up in other people's threads, in their audiences, where attention already lived. There's a reason for that: in the algorithm, a reply is worth far more than a like. Engagement that pulls you into someone else's conversation is the lever, and I had been pointing my product at the wrong lever entirely.
That reframed everything. Creating content was never the hard part. Getting anyone to see it is the hard part. Every tool I was competing against, mine included, was optimized for creation - the easy half - and ignored distribution, the half that actually decides whether you grow. That gap is the reason to keep building. I went deeper on it in my comparison of the best tools for creators.
Everything I got wrong (so far)
"I made mistakes" is useless. Here are the specific ones.
The brand voice system was a disaster at first. Version one read your posts and produced a weirdly formal version of you, like someone wrote a cover letter in your style. Same vocabulary, completely wrong vibe. Fixing it meant getting serious about which model actually captures a voice, so I ran a real bake-off, putting GPT, DeepSeek, GLM, and Claude head to head on the same brand. The differences were not subtle. That work is what turned brand voice from a gimmick into the part people stay for.
I built two months of analytics nobody wanted. Engagement broken out by day of week, 30-day rolling averages, beautiful charts. No one asked for it. Creators want to know "is this working," not read a spreadsheet. The analytics are simpler and more useful now.
I built Autopilot before I understood the trap inside it. Letting AI write and post in your voice sounds like the finish line. It isn't. A model will hand you a caption that is grammatically perfect, on-brand, and completely dead - confident and flat at the same time, the kind of post that reads fine and lands on nobody. Learning to tell a competent post from a post that will actually move was harder than building the feature. It is the difference between automation that saves you time and automation that quietly makes you forgettable.
Building from Romania
Software doesn't care where it was written. The context still shapes the build.
The obvious part is cost. My burn rate is a fraction of what it would be in San Francisco - no office, no payroll, no kombucha line item. Bootstrapped, no funding, a team of one.
The less obvious part is isolation. There is no co-working space full of founders here, no meetup where someone says "you should talk to my friend at Y Combinator." When I tell people locally that I'm building a SaaS, most of them ask what SaaS stands for. My accountant asked me to explain recurring revenue. Twice.
There's a line that stuck with me: the product with a sizable market and weak competition wins even with bad marketing, but in the same market the product with better marketing wins every time. After years of believing code quality was the whole game, that one stung. Building somewhere nobody around you understands the work makes the marketing gap wider, not smaller. The indie community online helps. Most days it is still you and a laptop.
What's next, and what I refuse to fake
The next real milestone is the loop. Right now Sydium writes in your voice and posts. It does not yet learn from what performed and adjust the next batch. Closing that loop - write, post, measure, adjust, repeat - is the line between convenient and genuinely useful, and it is where most tools stop because the measure-and-adjust part is hard.
More platforms come after that. Bluesky, Pinterest, YouTube. Each one is a week of OAuth headaches and API docs written by someone who hates developers, and each one is a reason to pick Sydium over a tool that covers three networks and stops.
What I won't do is fake the building-in-public part. No hockey stick I don't have. No revenue screenshot with a rocket emoji. I have a product I use every day, a small group of early users who actually tell me what's broken, a to-do list longer than I'd like, and the genuinely strange feeling of building a social media tool while still being mediocre at social media.
That last part is the point, though. For most of my career the people my software served were abstractions. Now the people who use Sydium talk to me directly about what they need, and the problem I'm solving is one I have. That makes it worth the bad days.
If social media is eating your hours, you can try Sydium for free - no credit card. If you're a fellow founder, I'm on Twitter. And if you think building another social media tool is a bad idea, you might be right. I intend to find out.