Google quietly drops an offline AI dictation app for iPhone

Dictating to your phone usually means ending up with a mess of "ums," "ahs," and half-baked sentences.

Google just quietly launched an offline-first AI dictation app called Google AI Edge Eloquent on iOS, taking direct aim at tools like Wispr Flow and SuperWhisper.

It is completely free to download, and in a surprising move, Google dropped it on Apple's platform first.

Gemma on-device, Gemini in the cloud

Eloquent is designed to work without an internet connection. Once you download its on-device speech recognition models powered by Gemma, transcription happens entirely on your phone.

As you speak, the app transcribes your words in real time. The moment you hit pause, it automatically filters out filler words like "um" and "ah" while smoothing out mid-sentence stumbles.

Underneath the transcript, handy quick actions let you reformat the text into bullet points, make it formal, or toggle between short and long versions.

If you prefer heavier editing, you can switch on cloud mode, which routes text through Gemini models for deeper polish.

To avoid butchering specialized jargon, the app lets you import custom keywords and names from your Gmail account, alongside a manual dictionary you can build yourself.

Each session tracks your stats, showing your total words, words-per-minute speed, and a searchable history of past dictations.

The disappearing Android mention

Right now, Eloquent exists only as a standalone iOS app. However, the initial App Store listing explicitly mentioned an upcoming Android version with system-wide default keyboard access and a floating widget.

Shortly after launch, Google updated the listing to remove all references to Android. Instead, it added a note confirming that an iOS keyboard extension is coming soon.

That means iPhone users will soon be able to dictate using Eloquent across any text field, not just inside the app.

Why it matters

We think this release signals real trouble for the crop of paid AI dictation utilities popping up lately.

Apps like Wispr Flow have built solid followings by offering clean voice-to-text with keyboard overlays, but they often require recurring subscriptions. Google is delivering on-device Gemma models and cloud Gemini cleanup for zero dollars.

For remote workers, writers, and anyone who thinks out loud while walking, offline processing eliminates both latency and privacy concerns about voice data leaving the phone.

The real question is how long standalone startups can survive once platform owners embed these capabilities directly into the keyboard layer. If Google ships the iOS keyboard soon and eventually bakes this natively into Android, the standalone dictation app market might shrink overnight.