Nikolay and I built thoughtbubble in PHP, JavaScript, and MySQL. There was no framework, no CMS, no boilerplate — we'd hit wall after wall and just keep watching web development videos until it clicked. I built the full back-end of thoughtbubble and co-developed the front-end, while also designing the full icon set and logo in Adobe Illustrator.

I built the back-end systems below, and co-developed the front-end and social feed logic with Nikolay:

• Authentication and session management
• News feed with custom sorting logic
• Post types — text, photo, video
• Likes, pins, and comments
• Hashtags and categories
• Notifications
• Profile customisation and account settings
• A user feedback pipeline

We shipped 76 major updates between the closed alpha and public launch.

The infrastructure ran on shared hosting with a manually configured MySQL database. There was no version control beyond file backups, no CI, no staging environment. It was scrappy by any standard, but it worked.

We scaled to nearly a thousand users, and every line of it was ours. Every feature or bug fix was a new learning experience and a sleepless night.

The pitch was simple: a social network where your feed only teaches you things. No vacation photos, no cat videos — just insights from books, actionable quotes, and self-improvement content. We built it in 2014, inspired by the rise of Gary Vaynerchuk and Tai Lopez and the audiences forming around them. People were already sharing nuggets from books, videos, and articles — but existing platforms weren't built to hold that kind of content well. We saw a clear gap: entertainment-first platforms simply weren't built to support information-rich content.

So we set out to build something different. We envisioned posts where the UI would include things like book titles, author, and reference material. A category system built around four pillars of self-improvement: health, wealth, relationships, and fulfillment. The user profile would read more like a personal development journal — with pinned posts, categorized content, and a space for reflection.

What we were describing — a curated, knowledge-first feed with social mechanics — is close to what Twitter's long-form threads, Substack, and LinkedIn's creator-focused pivot eventually became, three years after we shipped ours. We were two teenagers with no funding and no network, so the timing didn't matter in the end. But the product instinct was right.

Looking at how much that space has grown since, there's real satisfaction in knowing two teenagers with no prior experience, no funding, and no network saw it coming three years early. I sometimes wonder what we could have built with even a fraction of the resources those platforms eventually had — but that experience taught me that the desired outcome isn't always the best outcome. There's a deeper plan at work, and life had other things in store beyond scaling thoughtbubble to its full potential.

We didn't run ads. We didn't have a budget. The growth strategy was entirely manual: reach out directly to people who were already creating self-improvement content on YouTube or Instagram, show them the platform, and get them to seed it with their content. We identified early that a knowledge-first feed is only as good as the content in it — if the first 50 users post junk, the product is dead on arrival.

We tracked registrations, session counts, and engagement metrics in MySQL and used those numbers to prioritise what to build next. The most-used features got more polish. The least-used ones got cut or rethought. We didn't call it a data-driven product loop at the time, but that's exactly what it was.

At some point, two teenagers decided their app needed real legal protection — because apparently "two guys with a login page" doesn't hold up as a legal entity on its own. So we hired a copyright lawyer, nodded confidently at documents we barely understood, and paid real money to learn what a retainer is the hard way. He did the bare minimum, billed like he'd written us a constitution, and left us wondering if we'd just been the world's youngest, most polite scam victims.

Classic story. We pushed through it anyway, filed everything properly ourselves where we could, and got the copyright registered. It cost more than it should have and took longer than it should have — but it's ours, on paper, forever, which is more than I can say for the money we spent on that lawyer.

In hindsight, it was my first real lesson in the unglamorous side of building something — the admin, the fine print, the people who see inexperience as an opportunity rather than something to protect. Nobody warns you that "founding a startup" sometimes means arguing about an invoice with a man in a blazer. I'd handle it very differently now. But we got the copyright — and honestly, that's a better ending than most first-lawyer stories get.