RSS Amplifier

AI Realist · Aug 12, 2026

What Two Weeks at AI Realist Taught Me About AI, Design, and Building Things

0
Sign in to vote or save

nichearizonatea · AI Realist

Taissiya is 16 years old and a high school student. One day, her mother asked me whether she could do an internship at AI Realist — a mandatory part of her school curriculum. I always have things to do at AI Realist, and I loved the idea.

I, Maria, see the new generation currently in school as people who will eventually be AI-native. They grew up with this technology the same way we grew up with the internet. And I have to say, I was not wrong here. I have taught many people how to use AI — mostly mid-career or senior employees across various fields, from engineering to finance. And I can tell you that among all the people who had never coded before, Taissiya learned it far faster than anyone I have taught.

I believe the new generation entering the workforce soon will have entirely different levels of productivity and approaches to problem solving, and honestly, I am excited to live through it.

Now, read Taissiya’s report about her internship.

When I started my internship at AI Realist, I expected to help with a few promotional images, make some Canva designs, and maybe support with small website tasks.

Two weeks later, I had worked on website pages, GitHub files, Cloudflare deployments, social media graphics, video covers, promotional videos, AI-generated images, and interactive web elements.

It was much more technical, creative, and fast-paced than I expected.

It was also a lot of fun.

A large part of my internship involved creating promotional content for AI Realist.

I designed visuals for articles, livestreams, discounts, social media posts, video covers, and website pages. I worked with different formats for LinkedIn, Substack, stories, banners, and videos.

At first, I thought making a promo would be simple: choose an image, add some text, and make it look good.

I quickly learned that it is rarely that easy.

A visual can have the right information and still not work. It can contain too much text, use the wrong colors, feel too much like an advertisement, or simply not match the personality of the brand.

Many of the first versions needed to be simplified. We would remove text, change the layout, adjust the image, create more space, or completely rethink the idea.

That process taught me that good design is often about knowing what to remove.

I also spent a lot of time working on the AI Realist website:

https://airealist.org/

This was one of the most interesting parts of the internship because I was not only designing how a page should look. I was editing actual HTML, CSS, and JavaScript files, uploading changes through GitHub, and checking whether they appeared correctly after being deployed through Cloudflare.

I worked on pages such as About, Press, Media Kit, and Services.

This taught me how precise website work needs to be. Every page has its own file. Every link has to point to the correct place. A small mistake can make the wrong page appear or stop an element from working.

Sometimes I changed the code and expected to see the result immediately, but the old version was still visible because the new deployment had not finished. At other times, one small piece of incorrect code affected the layout of the entire page.

At first, this was frustrating. But once I started understanding the structure, it became satisfying. I could see how the website was built, identify what had gone wrong, fix it, and then watch the corrected version appear online.

Before the internship, I did not expect to work so much with code.

AI tools made it possible to create complex website elements very quickly. I used them to help generate HTML, CSS, and JavaScript for menus, interactive cards, filters, hover effects, animations, search functions, and responsive layouts.

One of my favorite ideas was a radar-style visual for “scanning AI hype.” It felt very connected to the AI Realist brand: instead of presenting AI as something magical, the idea was to look through the hype and find the useful information.

But I also learned that AI-generated code is not automatically correct.

AI can produce an impressive amount of code in seconds, but you still need to understand what the page is supposed to do. A feature can look interesting without being useful. An animation can work technically but distract from the message. Code can also be placed in the wrong file or interfere with something that was already working.

AI made coding faster, but checking, testing, and correcting the result was still essential.

The biggest surprise was how much human judgment is still needed when working with AI.

From the outside, content creation with AI can look almost automatic. You write a prompt, generate an image, place it into Canva, and the content is finished.

In reality, generating something is usually the easiest part.

The difficult part is deciding whether it is good.

Does the visual fit the brand? Is the message clear? Is there too much text? Does the design feel premium or generic? Would someone stop scrolling when they see it? Does it communicate one strong idea, or is it trying to say too many things at once?

AI can generate many options very quickly. That means the person using it needs to make even more decisions.

You have to choose what to keep, what to change, and what to remove.

I also became much better at prompting.

At the beginning, it was tempting to ask AI to “make a cool promo” or “create an AI image.” The results were usually generic.

The output became much better when the instructions became more specific.

Instead of asking for a general AI visual, I learned to describe the brand, platform, audience, colors, mood, layout, and what should not appear in the image.

For example:

“Create a premium editorial-style LinkedIn visual for an AI analysis brand, using cream, coral, brown, and purple, with one strong metaphor and very little text.”

That gives the AI much clearer creative direction.

I learned that prompting is not only about telling a tool what to generate. It is about explaining the idea behind the result.

Further in the article: how to use different tools together, learn to accept feedback, and the biggest takeaway.

All proceeds from paid subscriptions to this article will go to Taissiya.

Currently, all yearly subscriptions are 20% off.

Read the original on msukhareva.substack.com

Comments

Nothing yet. Say the first thing.

    Sign in to join the conversation.