About the Project
Since my last post about the program, I finally made it to the end of the journey. At the final stage, there was a competition. Unfortunately, I did not make it as one of the winners.
Yeah, that part sucked a little.
But this project is still something I am proud of, because it pushed me to build fast, think deeper, and turn an idea into a working AI product in a very short time.
The challenge was quite clear:
- create a video
- build an AI application using Qwen
And honestly, this is where I made my mistake.
I spent way more energy building the application than preparing the video. We only had one week, and I probably should have built something simpler. Instead, I overthought the concept and went all in.
Rather than making a small suggestion tool, I tried to build something closer to an AI "spaces" platform.
And yes... I ran out of time. I did not even finish the video properly.
Still, the app itself became a really interesting experiment.
What is Q-WanFlow?
Q-WanFlow is a platform inspired by products like Freepik Spaces, but focused on exploring Alibaba AI models such as Qwen and WAN.
The idea is simple: instead of forcing users to test everything through raw API calls or Postman, Q-WanFlow gives them a more practical interface where they can plug in their own API key and start generating content directly.
With Q-WanFlow, users can explore AI generation workflows such as:
- image generation
- video generation
- model experimentation
- prompt-based creative workflows
The goal was to make Alibaba's model ecosystem feel more usable for normal builders, not only for people who are comfortable testing APIs manually.
Why I Built It
This project came from a very simple problem.
Alibaba has powerful AI models, but at the moment there is still no widely adopted ready-to-use app experience for exploring WAN/Qwen in a smooth way. Most of the interaction is still API-first.
And honestly, opening Postman every time you want to generate something is not fun.
So I thought:
Why not build a lightweight platform that makes these models easier to try, faster to use, and more accessible for creators and developers?
That became the foundation of Q-WanFlow.
The Main Idea
Q-WanFlow is designed as a playground where users bring their own Alibaba API key and use the platform as a frontend experience for generation.
That means users do not need to depend on a separate subscription-based app just to experiment. They can use their own key, test prompts, generate outputs, and explore what the models can do in a cleaner workflow.
In short, the project tries to bridge the gap between:
- powerful AI APIs
- practical product experience
Tech Stack
This project was built with a simple but effective stack:
- React for the frontend UI
- FastAPI for the backend API layer
- Tailwind CSS for styling
FastAPI played an important role here because I wanted a backend that was lightweight, fast, and comfortable for handling AI-related request flows.
What I Learned
This project taught me a lot in just one week.
1. Shipping fast is hard
When the deadline is short, scope matters more than ambition. I learned that a smaller finished product is often better than a bigger unfinished one.
2. Product thinking matters
This was not only about calling an AI API. It was about creating a usable experience around the model.
3. AI tools need good UX
Even if the underlying model is powerful, people still need a simple interface, clear flow, and fast feedback to enjoy using it.
4. I enjoy building AI products
This project confirmed that I really like the intersection of engineering, product thinking, and generative AI.
Try It Out
You can check the live project here:
Live demo: q-wanflow.my.id
The repository is also public, so feel free to clone it and explore the implementation yourself:
Repository: View the repo
If you already have an Alibaba API key, feel free to try it out.
Final Thoughts
Even though I did not win the competition, I still see Q-WanFlow as a win for my own journey.
It started as a competition project, but it became something bigger: proof that I can take a raw AI capability, wrap it in a better product experience, and build something real under pressure.
Built with a mix of vibe-coding energy... and a bit of my own brain too.
Huge thanks to CBNCloud and Alibaba Cloud for the opportunity and the learning experience.
If you want to see more about the story and the video version, I will also share it through my .