- A “Forced True-香” Experience The moment that truly changed my opinion about voice input didn’t happen at my desk — it happened in the car.
One time, I was rushing to wrap up a feature request that wasn’t yet finished — something like “We need to redo the OAuth login flow”:
The old solution had historical baggage
The new requirement kept adding conditions
And in between were buried tracking points, pitfalls, and tons of compatibility details
The problem was: the car was already on the road.
It was bumpy, and I already get carsick
My posture was awkward — my arm would get sore after being stretched out for a while
The lighting and screen angle were uncomfortable
In short — this was an extremely unfriendly environment for typing.
But I didn’t want to waste that time, so I reluctantly tried voice input again for some “Vibe Coding”:
First, I went over the background of the current requirement
Then I added all the pitfalls I had run into before and compatibility considerations
I kept overturning my own ideas mid-way: “No, this would mess up the login state.” / “Wait, then the tracking data wouldn’t align.”
If it had been my past impression of voice input, the end result would almost certainly have been:
A pile of misrecognized text
A messy draft that I’d have to clean up myself
Here’s a smooth, natural English translation of your follow-up text, keeping the reflective and slightly conversational tone intact:
The results that time, collaborating with an AI Agent, turned out better than I expected:
It basically understood what I was saying
It organized all those hesitant, back-and-forth thoughts into a few reasonably clear solution options
It even put together a checklist of “pitfalls to watch out for during the redesign”
Sitting in that swaying car, talking while reading its responses, I suddenly had a very clear feeling:
It wasn’t just “typing” for me — it was thinking with me.
By the time that day wrapped up, a thought popped into my head: “Maybe this thing isn’t as useless as I thought.”
When the car finally arrived and I got back to my desk, I kept using voice input — but now with a question in mind:
If, in such a terrible environment for typing, voice + AI was still enough to get this feature done… did that mean my old conclusion — “voice input is useless” / “not suited for real work” — was simply wrong?
From that point on, I went from using voice input occasionally to making it a high-frequency habit (even upgraded my gear for it). It’s been over a month now.
And after more than a month of practicing voice + AI coding, my view of it has shifted dramatically. In this piece about voice input, I want to share three things:
Why I used to think voice input was a waste of time
What exactly changed to make me “forced-true香” about it
In AI coding, what voice input is good for — and not so good for
Top comments (0)