You must know that the app’s “microphone icon” thing where you can talk and it turns it into the text you would have had to type, is constantly messing up and needing to “retry” transcribing for whatever reason. Sometimes it is 10-20% of the transcriptions, and sometimes it is like 70% where most just need to be resent a few times using the “circling arrow icon” that is shown when it runs into an error.
I was excited to see the new feature you have but found it really sucks in the current form.
No retry available even though it still fails a decent amount of the time
Uploads the transcription right away, preventing you from being able to manually add stuff and correct the many, many typos and transcription mistakes it makes
I am a big fan of your product, but if I have a 90 second transcription fail and not be able to resend, I will go use ANY OTHER AI THAT DOESN"T WASTE MY TIME. Time is valuable, you are supposed to be saving it, so when you don’t do that, you are not worth paying for. It won’t kill you to add the retry icon back, but until then I’m going to be trying out other AI’s.
By the way, as a developer, whoever is doing your app testing really needs to expand their edge case testing and think though the end-to-end use of their product. Your past few app releases are just sloppily done.
I’m here to second this. I actually thought the old t2t did a good job. The new update is horrendous. Can’t add images. Automatically sends the text without allowing any manual addition to it. The epitomy of breaking something that wasnt broken. You took a good product, made it worse, and called it an update.
This “auto-send” voice-to-text feature is a complete UX failure.
The fact that I can’t even review what I just dictated before it sends is beyond stupid, especially in a platform built on thoughtful, precise language. There was a short window where it reverted back to the old system, and it worked flawlessly. Then, with no warning or option, you ripped it away again.
This isn’t innovation. This is regression.
Not every message should be fired off the second I stop talking. Sometimes I need to breathe, edit, rephrase, or cancel. That’s how real conversations work, especially with an AI where wording matters. I shouldn’t have to race to hit stop, copy, and restart just to avoid sending something premature or half-baked.
If this is A/B testing, then it’s failed. If it’s intentional, then it’s arrogant. And if it’s a temporary rollback from a better version, then your priorities are seriously messed up.
Give us a toggle. Let us decide whether messages send automatically or not. You made it worse and offered no way back. That’s not how good software evolves. That’s how it devolves.