Willow on X

The largest group is plain enthusiasm: 23 people say Willow is the best speech-to-text they have come across — "the goat", "literally the best" — or recommend it to whoever asks for a dictation tool. What they name is speed and not touching the keyboard: one uses it toward inbox zero, one has the action button on their phone write emails, another likes that it takes over caps lock.

9 people praise Wally for keeping the conversation going until the text is right and for learning a person's voice; one says a brain dump comes back as clear writing, and one reports that the post itself was written with Wally.

Below those, 6 people say they picked Willow over Wispr Flow, some having switched or cancelled months earlier. 6 are critical — repeated crashes, no dark mode, a free tier they call misleading. 2 are still weighing the two.

The largest complaint is reliability: 6 people report the app crashing, running slowly, or breaking a key. One counted eight crashes during a homework assignment and went back to another tool; others say it is slow in APAC, that a single key modifier no longer starts dictation on Windows, that it does not work in the Copilot app, that weekly usage stays stuck at zero, and one asks Willow and Granola to get along on the same Mac.

3 people say the free plan is too limited to be worth using — that calling it free is misleading, or that it is barely usable without smart controls — and one notes a $15/month in-app purchase on iOS.

Three more complaints sit level at two people each: no dark mode, with the white screen that "blows me up"; transcription and AI formatting less accurate than a rival's; and no Android or Linux version.

Two people accuse the blog of false claims about a competitor — a small theme, but its post is the furthest-reaching in this section.

Wispr Flow is the rival to beat. 15 people compare the two, and the split is real. Some have switched to Willow and never looked back, one cancelled Wispr, and one says Wispr "cares way too much about marketing and not the actual product"; others credit Willow's free version as accurate against Wispr's paid one. Going the other way, one person who tried to switch says Willow is "nowhere near good enough" — a buggy iOS app, worse accuracy and worse AI formatting — and went back.

One points out that typing "wi…" surfaces Willow ahead of Wispr, and a reply asks whether the two are just the same. Another still cannot choose between them.

The rest is thin, one person each: someone weighs Stirae and cites latency under 40 ms; someone asks whether Willow beats the free ChatGPT dictation on Mac; someone frustrated by the missing dark mode would settle for whatever dictation the platform offers; someone just wants Willow and Granola to coexist on a Mac.

The largest group is people waiting on the switch offer. 4 say they filled in the form and sent what was requested, then heard nothing — one asks whether they "accidentally submitted my application on mute", another whether the offer is still ongoing. Nobody reports a reply.

2 people ask about Android: one is "Waiting for Android version", another asks whether anything has shipped yet.

Two further questions are level at two people each. One is how Willow compares before deciding — whether it is better than the free ChatGPT dictation on Mac, and what exactly is better about it than Wispr Flow. The other is what the free plan actually includes: whether the paid plan runs a different, more accurate model, and whether free really means free when a $15/month in-app purchase shows on the iOS app.

This is a short section, 5 people, and the largest group is 2 people waiting on an Android app. Neither of them describes anything more than that.

The remaining three are level, one person each. One wants long dictation to drop what the speaker corrected — turning "Tuesday—actually, Wednesday" into just "Wednesday", to save cleanup after the text lands. One wants emojis inferred from the audio, laughs and tone especially, optionally on WhatsApp, so they need not add them by hand after transcribing. One would rather features rolled out plainly: "Just roll out the feature without the cringe mascot / naming thing."

Two of those asks, one person each, are about the output arriving finished — corrections already dropped, emoji already in — rather than about the dictation itself.