HomeAboutTimelineCVProjectsYouTubeBlogArticlesBooksContactBook a Call
Back to Blog
aiside-projectno-codebuild-in-public

I Built a Thing Called Emojar — With Zero Lines of Code

I shipped a complete web product — frontend, backend, SEO, content, deployment — without writing a single line of code myself. Here's what that experience taught me about AI in 2026.

May 14, 20264 min read

I Built a Thing Called Emojar — With Zero Lines of Code

I built a thing. 🫙

Meet Emojar.com — an emoji search and copy tool. Nothing revolutionary. Nothing that's going to disrupt the industry. Just all the emojis in one jar, a clean search interface, some curated collections, and a few mini games for good measure.

The fun part? I wrote zero lines of code.

Not "almost zero." Not "I only wrote the config." Literally zero. The entire product — frontend, backend, SEO, content, deployment — was orchestrated through AI tools. I played product manager to a team of robots. And honestly? They delivered.

What Is Emojar?

Emojar is a straightforward emoji utility:

  • 3,600+ emojis — searchable and copy-paste ready
  • Categories and A–Z browsing — so you can find what you need without guessing the name
  • Curated collections — hand-picked sets for specific moods and use cases
  • A blog — yes, the robots write too
  • Emoji mini-games — because why not 🎮

It's the kind of small tool I've always wished existed in a cleaner form. Pick an emoji, copy it, move on. No fuss.

The Experiment

This wasn't a project I planned extensively. It was a weekend experiment in one specific question:

How far has the AI ecosystem actually come in 2026 — can it ship a complete, production-ready product end-to-end?

The answer, it turns out, is: pretty far.

I didn't write HTML. I didn't write CSS. I didn't configure a server or touch a deployment pipeline by hand. I described what I wanted, reviewed outputs, gave feedback, and iterated. The AI tools handled everything else.

Here's roughly how the workflow looked:

  1. Product definition — I described the core concept, the pages I wanted, and the user experience
  2. Frontend generation — components, layout, responsive design, dark mode, all AI-generated
  3. Data layer — emoji data structured and searchable
  4. SEO & content — meta tags, descriptions, even the blog posts
  5. Deployment — infrastructure configuration handled without a single manual step

What I Actually Did

Let me be precise about my role, because "zero lines of code" can be misleading.

I did a lot. Just not coding.

  • I made product decisions — what features to include, what to cut, what the UX should feel like
  • I wrote prompts — carefully, iteratively, with specificity about what I wanted
  • I reviewed outputs — not everything the AI produced was right; judgment still matters
  • I iterated — multiple rounds of feedback until things felt right
  • I curated — the emoji collections, the blog topics, the overall tone

This is the new shape of certain kinds of work. Not "AI replaces the human" — more like "AI compresses the skill surface required to build something." You still need taste, judgment, and a clear vision. You just no longer need the implementation knowledge to execute it.

What This Tells Me About 2026

A few honest observations:

The gap between idea and shipped product has collapsed for certain categories. Simple utilities, content sites, tools — these can now be built by someone with good product sense and no engineering background. That's a meaningful shift.

Prompting is a real skill. Getting AI tools to produce good, production-quality output requires clear thinking. Vague inputs produce vague outputs. The people who will get the most out of these tools are the ones who can think precisely about what they want.

Taste still matters — maybe more. When the cost to produce drops, the differentiator becomes judgment. Knowing what's good, what's not, what to cut, what to keep. That doesn't get automated away.

Some things are still hard. Novel architecture decisions, debugging subtle state bugs, performance optimization at scale — AI tools are much weaker here. Emojar is a simple product. I wouldn't try to build a distributed database this way.

Is It Going to Make Me Rich?

No.

Did I learn a ton about what AI can actually ship end-to-end in 2026? Absolutely.

That was the point. Emojar is a proof-of-concept disguised as a product. The real output wasn't the emoji search tool — it was a clearer picture of where the frontier actually is.

Go Try It

emojar.com — go find your favorite emoji.

And yes, I'm mildly proud that my most "complete" side project has zero lines of my own code in it. That says something about 2026.

If you want to follow more experiments like this, I write here and make videos on the SiaExplains YouTube channel about exactly this kind of thing — AI tools evaluated honestly, not hyped.

All blog posts