黑客松作品：打造語音智能體，探索更自然的軟件互動

This is a submission for the MLH x DEV Writing Challenge.

What I Built

During the hackathon, I worked on a voice agent — a project that explored making interactions with technology feel more natural through voice.

The idea was to move beyond the usual text-based interactions and explore a more conversational way for people to communicate with software. Building it meant figuring things out as I went, experimenting with the available tools, and trying to turn an idea into something that actually worked.

The experience was hectic, with a lot happening in a short amount of time. Looking back, what stands out isn't every technical detail but the process itself: making decisions under pressure, troubleshooting unexpected issues, and learning by building.

That experience is one of the reasons I keep coming back to hackathons. They push you to try things you might otherwise postpone and give you an excuse to turn an idea into a real project.

Demo

• Voice-agent-workshop : https://github.com/coder-irwin/voice-agent-workshop(we were told to do this in the offline meetup)
• ChibiDesk: Microsoft Store listing
• ChibiDesk source and releases: https://github.com/gen-ayu/ChibiDesk-Releases
• ChibiBot repository: https://github.com/gen-ayu/ChibiBot

Partner Technologies

The project explored voice-based interaction and the possibilities of building a more conversational software experience.

Working on a voice agent involves thinking about how users express their requests, how software processes those requests, and how responses can make an interaction feel natural rather than mechanical.

I'd like to revisit the project with a clearer picture of its architecture and continue exploring how AI can make voice interfaces more useful and intuitive.

Hackathon Experience

That week was a blur of building, experimenting, and trying to get everything working within the time available.

Hackathons have a particular kind of energy: you start with an idea, discover problems you didn't anticipate, and suddenly find yourself making decisions you would normally spend much longer thinking about.

It wasn't a perfectly organized development process, and that was part of the experience. Sometimes you have to prioritize getting something working over making everything perfect.

I might not remember every detail of that week, but I remember the feeling of trying to build something meaningful under a deadline. More importantly, it reminded me that the best way to learn a new technology is often to stop waiting until you know everything and start building.

That's a mindset I've carried into my other projects ever since.