Why I'm Building a Social Media Tool After 15 Years of Coding

A Decade and a Half Building Software I Didn't Own

I started coding at 21 in Romania. Mobile apps, games, whatever I could get my hands on. Some genuinely fun years - I learned how to ship, how to handle users, how to build from nothing. Then somewhere around year 10 something shifted. The projects stopped mattering. Features I thought were pointless, architecture choices that made me cringe, sprint planning meetings where we spent an hour debating button colors. All of it compounding into one question: why am I doing this for someone else?

I didn't have a dramatic exit. No quitting story. I just started building Sydium on nights and weekends like every other solo founder who can't afford to quit their job to see if anyone wants what they're making.

"Aren't There 50 of These Already?"

Yes. And the social media management market is worth $32 billion, growing at 24% per year. A crowded market doesn't mean "go away." It means people are paying and the existing tools got comfortable.

I researched all of them. Buffer, Hootsuite, Later, Metricool, Sprout Social, Publer. Every time the same pattern: the tool does 70% of what creators need and makes the other 30% painful. Or it does everything but costs $249/month, which is more than my rent in Romania.

Here's the specific gap I saw. Buffer was revolutionary in 2013. Hootsuite was the answer when "social media manager" became a real job. But both were built before AI could write a decent paragraph. Before one person could realistically manage six platforms with the right software. When AI showed up, most of them just bolted a "generate caption" button onto the same scheduling interface from 2015 and called it innovation.

I wanted AI to be the foundation, not a feature. Sydium learns how you write - your vocabulary, your tone, your weird emoji habits. Then it creates content that sounds like you, schedules it, and publishes. You review once a week or let it run on autopilot.

Everything I Got Wrong (So Far)

Let me be specific because "I made mistakes" is vague and useless.

Mistake 1: The brand voice system was a disaster. The first version analyzed your posts and produced this weirdly formal version of you. Like if someone read your tweets and then wrote a cover letter in your style. Technically similar vocabulary, completely wrong vibe. That took three months to fix. It's still getting better every week.

Mistake 2: Two months on analytics nobody wanted. I built a dashboard with engagement rates broken down by day of week, 30-day rolling averages, beautiful charts. Nobody asked for it. Creators want to know "is this working" - not stare at a spreadsheet. The analytics are simpler now, more actually useful.

Mistake 3: Ignoring distribution entirely. This was the big one. I built a great tool for creating and scheduling content. But creating and scheduling was never the hard part. The hard part is getting anyone to actually see what you made. My own Twitter data proved it. The 332K impressions came almost entirely from replies, not from content I published. 70% of the battle is distribution. 30% is creation. And every tool I was competing with - including my own - was optimized for the 30%. That realization changed the entire product direction. Not just publishing. Something that understands what performs and adjusts over time.

The Real Numbers

I'm going to share specific numbers because most "building in public" posts don't.

Twitter (4 months in):

• Peak: 332,000 weekly impressions
• Profile visit to follow rate: 7-8% (92% of people bounce)
• Best single reply: 1,300+ likes
• Average original post: under 100 impressions
• Total ad spend: $0

The business:

• Users: small but growing (not thousands, not going to fake it)
• Funding: $0 raised, bootstrapped from Romania
• Team: just me
• Monthly costs: Firebase hosting, AI API calls, a domain name

What I learned from those numbers: replies outperform original content by roughly 10x on Twitter. One cynical one-liner on a viral thread gets more visibility than a week of polished posts. The 60/30/10 ratio that works for me is 60% visibility replies on big threads, 30% hybrid replies that reveal who I am, 10% warm replies to small builders. The hybrid replies are where follows actually come from.

Building from Romania

Software doesn't care where you wrote it. But the context shapes things.

The obvious part is cost. My burn rate is a fraction of what it'd be in San Francisco. No office, no employees, no kombucha budget.

The less obvious part is isolation. There's no co-working space full of founders here. No meetups 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 quote that haunts me: "The product with a sizable market and low competition wins even with bad marketing. But in the same market, the product with better marketing wins every time." After all those years thinking code quality was everything, that one stung. Building from a place where nobody around you understands what you're building makes the marketing gap even wider. The indie hacker community is online, which helps. But it's still you and a laptop most days.

What's Next

I'm building in public but not the "day 47 of my journey" kind that's mostly content marketing dressed up as transparency.

• The AI feedback loop. Right now Sydium creates content in your voice and posts it. But it doesn't yet learn from what performed and adjust the next batch. That's the gap. When it can close that loop - write, post, measure, adjust, repeat - that's when it goes from convenient to genuinely useful.
• More platforms. Bluesky, Pinterest, YouTube. Every new platform is a week of OAuth headaches and API documentation that was clearly written by someone who hates developers. But each one is a reason someone picks Sydium over a tool that only covers three or four networks.
• Better distribution tools. Because the data is clear - creation is 30% of the problem and distribution is 70%. Most tools ignore the 70%.

The Honest Version

I don't have thousands of users. No hockey stick chart. No revenue screenshot with a rocket emoji. I have a product I use every day. A small group of early users who actually give feedback. A list of things to build that's way longer than I'd like. And the weird contradictory feeling of building a social media tool while still being mediocre at social media.

But I've spent 15 years building software that solved problems I didn't care about for people I'd never meet. Sydium solves a problem I have. The people who use it actually talk to me about what they need. That alone makes it worth it even on the bad days.