跳到正文
DEV Community· Garon·· 12 天前AI 評分31

複製用戶期待的功能,但打造自己真正需要的東西

Copy the expected, but build what you need.

AI 導讀

Garon 分享打造聽寫及會議應用 Glimpse 的經驗:先加入用戶期待的功能,再從實際使用中找出值得改進之處。他受課堂筆記需求啟發,打造能配合即時逐字稿更新筆記的功能,並指出持續使用自己的產品有助及早發現問題。

正文

Garon

Garon

Posted on Nov 10 AI-assisted

For the last year I've been working on an app called Glimpse. It's a dictation app and a meeting app, built pretty much all by myself. I want to share something I've learned about building apps like this, where people come in with expectations, and how to build something that sets you apart.

After spending a few months getting Glimpse to have every feature anyone would expect in a dictation app, I wanted to look at meetings, since they could be built on the same models and interfaces. One of my big inspirations there was MacWhisper. After I'd released a working meeting mode, I got an email from them about a new feature they'd launched called live transcripts. I thought it was cool, so I built it in.

The important part is that I used it. I noticed what I didn't like about it. Then I'd sit through lectures writing notes, and I'm thinking: I'm writing notes, but I also have live, diarized information coming in. Why can't a model use that?

I'd rather look at notes as the lecture happens than paste the whole transcript into a model at the end and have it write notes from that. This way I can work with the model. I write my own notes, it updates them, and everything in between.

So I built that, ran the development version of my app, and had it take notes through my lectures. I realized that sometimes it would listen too long or too short, and it wouldn't tell me which model was listening to my live notes.

A few things I'd pass on from this:

  • Copy what people expect. I built live transcripts because MacWhisper had them and people want them. There's no point reinventing that.
  • Pay attention to what you don't like, even in the stuff you copied. That's where the thing only you would build comes from.
  • Use it everywhere. For me that was sitting through lectures, not testing it for five minutes.

But the most important part of building anything is being obsessed with it and using it all the time. You'll notice the issues with what you built, and the odds that someone else will see those issues, or relate to them, are very high. So you fix them before other people ever try it.

If you build something you don't use, then other people probably won't use it either.

來源:DEV Community · dev.to