Software as Self-Expression

Since Gemini 3 dropped, I’ve been vibe coding a lot lately. A few of my demos went viral on X, and I don’t think it was because of any technical complexity. Most took just a few prompts. What resonated was the opinion baked into each product.

There’s something intimate about building software that we don’t talk about enough. When someone uses a product you’ve made, they’re seeing the world through your eyes.

I’ve always expressed opinions through content: blogs, podcasts, videos. Recently I discovered a new medium: software.

Two examples:

(577k views on X)

Most people freeze up when talking to a camera. It feels like shouting into the void. When I record, I imagine chatting with a friend. That’s what makes me natural on camera.

So I built something that talks back while you record. Real-time feedback from an AI that asks questions and responds. Suddenly it’s a conversation, not a monologue.

The product embeds an opinion: humans need dialogue to think clearly. AI shouldn’t just answer our questions. It should ask them.

As a variation of this idea, I also built another demo that simulates a live stream with real-time AI audience hyping you up as you’re recording a video.

(88k views on X, 340k views on RedNote)

Traditional dictionaries are dead artifacts. Published once, updated rarely. But language moves fast: new slang, memes, cultural references.

I wanted a dictionary that’s as alive as language itself. One that tells me why this Japanese politician is trending. One that explains “scroll” vs “march” with video. Context matters as much as definition.

The opinion: language isn’t just vocabulary. It’s culture. Your dictionary should be your cultural guide, not a textbook.

Throughout human history, our storytelling mediums keep evolving. Oral tradition became writing, became blogs, became video. Now it’s software.

Have a strong opinion? Don’t just tweet it. Build it into a product that makes people experience your worldview.

The beauty is that opinionated products distribute themselves. People share them because they agree with the underlying philosophy.

This is especially true for AI-era side projects. Building is so cheap now that “too niche” isn’t a criticism anymore. You’re not pitching VCs. You’re not quitting your job. So stop building for mass appeal.

Pick one weird problem for one specific type of person. Make something that would get killed in every corporate product review. Make something that only makes sense if you share a particular worldview.

Make software that sounds like you.