The Story Behind Kiolix Pulse

I have spent most of my career building software.

Over the years, I have worked on web services, mobile applications, backend systems, cloud infrastructure, and various other software projects. Building software was what I knew best.

The people around me often shared similar interests and backgrounds. At some point, I became curious about something outside of my usual circle.

What are people actually interested in?

I knew how to build software. I knew how to solve technical problems. But I realized I knew very little about what millions of people were searching for, talking about, or paying attention to every day.

That curiosity led me to Google Trends.

Collecting Google Trends Data

It started as a simple curiosity.

While exploring popular search terms in Korea, I began wondering what people in other countries were searching for.

What topics are trending in the United States?

What stories are people talking about in Japan?

What is happening across Europe, the Middle East, and Southeast Asia?

To answer these questions, I started collecting Google Trends data myself.

At first, I tracked trends from 10 countries.

Over time, the project expanded, and today Kiolix Pulse collects Google Trends data from 26 countries.

As more countries were added, another challenge emerged.

Many of the trends were written in languages I could not understand.

A trending keyword is not very useful if you have no idea what it means.

To make global trends easier to understand, I introduced multilingual support.

Initially, the service supported 11 languages:

• Korean

• English

• Japanese

• Simplified Chinese

• Traditional Chinese

• Spanish

• French

• Portuguese

• German

• Indonesian

• Arabic

The goal was simple: allow users to explore trends from around the world regardless of language barriers.

However, real-world usage taught me an important lesson.

Translation costs and AI token usage increased significantly, while user growth remained largely unchanged.

Eventually, I decided to simplify the service.

Today, Kiolix Pulse primarily supports Korean and English, while providing translation features that help users understand the meaning behind trending keywords.

Although translations are not always perfect, especially for certain regions such as Taiwan, they have proven extremely useful for exploring trends across different countries.

What Else Are People Watching?

As I collected search trends, new questions emerged.

What videos are people watching?

What products are they buying?

What movies and TV shows are gaining attention?

To answer those questions, I started collecting additional datasets beyond Google Trends.

I added YouTube trending data.

I collected shopping rankings from online marketplaces.

I even experimented with tracking OTT platform rankings.

Not every experiment succeeded.

While gathering OTT-related data, I eventually encountered IP blocking issues and had to remove the feature entirely.

That experience taught me something important.

Not every idea deserves a permanent place in a product.

Only the features that create real value survive.

Wanting More Than Just Trends

As I continued monitoring trends, one question kept coming back.

Why is this keyword trending?

At first, I manually searched for news articles and related information.

But as the number of trends grew, I found myself repeating the same research process over and over again.

To make things easier, I started attaching related news sources automatically.

I integrated Google News RSS feeds.

I added Wikipedia information.

I included movie and TV series references when relevant.

This helped me understand trends much faster.

However, there was still a problem.

Reading multiple articles, visiting different websites, and piecing together information remained a time-consuming process.

Using AI to Analyze Trends

Eventually, I asked myself a simple question.

What if AI could write a trend analysis article using the information I already collected?

That idea led to a series of experiments.

I provided trend data, news articles, and supporting information to AI models and asked them to explain why a topic was trending.

The first results were not very good.

But after continuously refining prompts and improving the information pipeline, the output gradually became more useful.

Eventually, I built a feature that allows users to receive AI-generated explanations without having to visit multiple websites themselves.

The goal was to answer the same question that started this project:

Why are people searching for this?

Keeping Server Costs Low

Kiolix Pulse did not begin as a business idea.

It started as a personal project driven by curiosity.

Because of that, I wanted to keep operating costs as low as possible.

In the beginning, data collection ran on an old MacBook.

As the project grew, I moved the infrastructure to AWS.

Data collection was handled through Python and JavaScript scripts, while the frontend was built with React.

To minimize costs, I chose a static website architecture using AWS S3 and CloudFront.

For the first few months, operating costs stayed around three dollars per month.

The dataset has grown significantly since then, but minimizing infrastructure costs remains an ongoing challenge.

A Series of Experiments

Along the way, I ran many experiments.

At one point, I built a Chrome extension because I wanted to see my collected data directly inside Google Trends.

The extension worked well until Google updated its website structure, forcing me to discontinue the project.

I also experimented heavily with SEO.

I created trend pages for search engines including Google, Naver, Bing, and Yandex.

Using more than 100,000 trend records, I attempted to maximize search engine indexing.

The results were humbling.

Search engines were not interested in pages that simply displayed raw data.

In particular, Google chose not to index most of those pages.

That experience taught me an important lesson.

Displaying data and creating value are two very different things.

Different Ways of Understanding the World

People consume information differently.

Some read news articles.

Some watch YouTube videos.

Some follow social media platforms.

Others spend their time in online communities.

Google Trends does not represent everyone's interests.

However, it remains one of the best ways to understand what large numbers of people are actively searching for and paying attention to.

Kiolix Pulse did not start with a grand business plan.

It began with a simple personal curiosity.

What are people interested in?

Why does a topic suddenly become popular?

I started this project to find answers to those questions.

And I am still looking for them today.