Joining a team at an AI hackathon has led me to try things I would never have done on my own!
This time, my new challenge was video production.
And not just any video. One with AI-generated narration in Vietnamese! (LOL)
I'd never even generated an AI voice in Japanese before, so I definitely didn't expect my first attempt to be in Vietnamese.
It took a lot of time, and in the end, we decided not to use the finished video for our final pitch.
But that's totally fine with me.
I'm actually really happy I had the chance to try something new.
Here's the story of my unexpected video-making adventure at Vietnam Japan AI Hackathon 2026.
Our Two-Person Team Made It into the Top 16!
In my previous article, I shared how I almost gave up after an orientation conducted in Vietnamese, but eventually found a teammate through English communication and formed a two-person team.
And guess what?
Our team, FrostLoop AI, made it into the Top 16!
We're developing an AI agent that tackles challenges related to selling on Amazon.
Since I actually sell products on Amazon, I'm contributing my practical experience in inventory management, business knowledge, and design skills.
My teammate is mainly handling the technical side, including AI agent development and backend engineering.
With Pitch Day coming up on October 24, preparations started getting serious.
And building the app wasn't the only thing we needed to do.
There were presentation slides, an introduction video, a project summary, and plenty of other things to prepare.
I had started working on some of the slides myself, but eventually my teammate took the lead in putting the final presentation together.
Meanwhile, I decided to take on video production!
Making a 47-Second Introduction Video with Canva
My original idea wasn't to explain every single feature of FrostLoop AI.
Instead, I wanted the video to make the problem we were addressing feel real and memorable.
The technical details could come later in the presentation. The video's job was to help the audience understand why the problem mattered in the first place.
So I opened Canva and started combining images, text, logos, music, and narration.
FrostLoop AI's visual identity uses icy shades of blue.
I also included our penguin character and screenshots of the actual app.
At first, I planned five different scenes.
But as I worked on the video, I started wondering, "Do I really need this many?"
Eventually, I simplified the structure to four scenes.
An opening, an introduction to the problem, a news report about the disposal of unsold products, and the solution FrostLoop AI aims to provide.
I used AI-generated images and added movement and music to make everything feel more like a video.
For the background music, I went with something a little unexpected: a hip-hop-style beat.
And you know how it goes.
Once you start editing, suddenly every little detail seems important! (LOL)
My First AI-Generated Voice — in Vietnamese!
The most exciting new experience for me was creating the narration.
I used Canva's AI voice generation feature to produce Vietnamese narration.
Here's the funny part.
I'd never generated an AI voice in Japanese before.
And suddenly, I was trying it in Vietnamese!
Since I don't speak Vietnamese, I couldn't properly judge the pronunciation or whether the translated narration sounded natural.
Still, once I added the voice to the visuals, the whole thing started feeling like a real presentation video.
But that's when things got a little tricky.
Adding narration means you need enough time for the words to be spoken.
I originally wanted to keep the video much shorter, but the more explanation I added, the longer it became.
I also wanted the background music to be quiet enough for people to hear the narration clearly.
That meant splitting the audio and adjusting the volume for different scenes.
Who knew there could be so much work behind such a short video?
The final version ended up being about 47 seconds long.
After spending quite a bit of time on it, I finally finished!
But We Decided Not to Use That Video! (LOL)
On October 18, I had a meeting with my teammate.
We discussed the 47-second video, too.
In the end, we decided it looked a little too AI-generated for the pitch, so we wouldn't use it.
After all those hours of work!
Of course, part of me thought, "Well, I did spend a lot of time making that..."
But I really wasn't upset about the decision.
When you're building something as a team, plans change.
Not everything you create will necessarily make it into the final project.
And I'm completely okay with that.
Because the experience of making the video is something I get to keep.
Generating Vietnamese AI narration, matching the visuals with the music, adjusting the timing of each scene — all of those were new experiences for me.
Honestly, if I'd joined the hackathon alone, I probably wouldn't have even thought about making a video.
So for me, having the opportunity to try something new mattered much more than whether the finished video was used.
And looking back, I can see why the video ended up feeling so AI-generated.
One reason was copyright. I couldn't use the actual footage from a television report in the UK.
The other was the language barrier. Because I don't speak Vietnamese, I relied on AI-generated narration and had no way to judge whether the translation sounded natural.
But my video-making adventure wasn't over yet!
This Time, I Was Going to Appear in the Video — Just a Little!
With our new approach, I would explain my own experience dealing with an Amazon-related problem.
Instead of an AI-generated character and AI narration, we would have a real person talking about a real experience.
And that person was me! (LOL)
On the evening of October 19, I recorded a new video using OBS Studio.
I displayed actual Amazon fee data on my computer screen, with a small webcam video of myself alongside it.
Then I explained the experience in Japanese.
The video included Amazon fees I had actually incurred in June 2026.
I'd struggled with this issue before, but I've since resolved it.
And to be clear, I didn't solve it using FrostLoop AI.
But having dealt with the problem myself, I know how time-consuming and complicated the decisions can be.
That real-life experience was also one of the starting points for our AI project.
Recording the video involved plenty of unfamiliar tasks.
I had to combine screen recording with my webcam, show the relevant data while speaking, and figure out how to make everything work together.
After recording, I used QuickTime to trim the footage down to about 20 seconds.
It was completely different from making the Canva video.
And just like that, I'd learned to use another new tool!
Vietnamese First, Japanese Second
As of October 20, the plan is to use my personal-experience video as Part II of the pitch introduction.
My teammate is currently working on Part I.
The idea is to introduce the broader Amazon selling problem first, then show my video as someone who has actually experienced it.
The pitch itself will be conducted mainly in Vietnamese.
But for my part of the video, I decided to speak Japanese — following my teammate's recommendation.
I actually asked, "Wouldn't English be easier for more people to understand?"
But my teammate thought the opposite!
That surprised me!
Any translation or subtitles will be part of the final preparations.
And after the video, I'm planning to introduce myself in Vietnamese for about 10 seconds!
My teammate will then introduce himself in Vietnamese and explain the project using our presentation slides.
Who would have thought I'd end up introducing myself in Vietnamese?
Although, to be fair, I'm the one who suggested doing it! (LOL)
I'm practicing now.
And I discovered that Google Translate can actually help me practice the pronunciation.
At first, I thought I'd need to ask my teammate to read the lines aloud and record them for me.
Turns out, I didn't need to!
None of this was something I imagined when I first signed up for the hackathon.
Working with people who know how to do things you don't can really expand what you're capable of.
Working with Others Leads You to Experiences You'd Never Choose on Your Own
What struck me most about this whole experience was how interesting it is to create something with other people.
Things I wouldn't have thought were necessary if I were working alone.
Tools I'd never used before.
Tasks I would never have chosen to do.
When you're working toward a shared goal with someone else, those things can suddenly become part of your role.
Of course, learning something new takes time.
And sometimes it's tiring.
I spent a lot of time making that 47-second video.
But I don't think any of that time was wasted.
Whether the finished work gets used or not, I had the chance to try something for the first time. And that, to me, is one of the rewards of joining a hackathon.
What's even better is that these experiences stay with me.
The next time I need to make a video, I'll already know how to create AI narration in Canva.
And I'll be much more comfortable recording my screen and webcam together using OBS Studio.
In this hackathon, my teammate and I are working on the same product from different directions.
He's handling the technical development, while I'm contributing business knowledge and practical experience.
I don't need to be able to do everything myself.
And not every idea or piece of work I contribute needs to be included in the final result.
In fact, working with someone else is exactly what opens up new possibilities for me.
And I think that's one of the biggest reasons I'm having so much fun participating in hackathons right now.
Pitch Day is coming up on October 24!
We're still making the final adjustments to the presentation materials and videos, but I'm really looking forward to seeing how everything comes together on the day!