Google has expanded its AI ecosystem on Apple devices with the quiet release of Google AI Edge Eloquent, a specialized AI dictation app designed specifically for iOS. Launched on Monday, April 7, 2026, the application aims to move beyond simple speech-to-text by using artificial intelligence to bridge the gap between raw, natural speech and professional-grade prose according to TechCrunch.
Unlike traditional dictation tools that capture every “um,” “uh,” and mid-sentence correction, Google AI Edge Eloquent focuses on capturing intended meaning. The app automatically filters out filler words and stumbles to produce clean text. This represents achieved through the use of Google’s Gemma architecture, utilizing efficient, open-weight, on-device models that allow the app to function without a constant server connection via the App Store.
The app is currently available for free on the iOS App Store, offering a subscription-free experience with no usage limits. By downloading Gemma-based automatic speech recognition (ASR) models directly to the iPhone, the software provides a responsive, offline-first experience that prioritizes both speed and user privacy as reported by Mashable.
On-Device Processing and Privacy Architecture
A central pillar of Google AI Edge Eloquent is its commitment to local processing. Most of the machine learning occurs entirely on the iOS device, meaning audio and personal data do not leave the hardware. This “zero cost architecture” is supported by Google AI Edge runtimes, which are optimized to provide high performance even on smartphones that are not the latest flagship models via the App Store.
While the app is designed for offline use, it does offer an optional “cloud mode.” When this feature is enabled, the app leverages cloud-based Gemini models to perform more advanced text cleanup and polishing. Users can toggle this mode off to ensure local-only processing for maximum privacy according to TechCrunch.
Customization and Contextual Accuracy
To improve dictation accuracy, the app includes a personal context dictionary. This editable dictionary learns from the user’s specific vocabulary to better recognize unique terms. For those who want a more seamless setup, the app can optionally connect to a Google account to build this dictionary using account data, which is then processed securely on the device via the App Store.
Users can too manually import keywords, names, and industry-specific jargon from their Gmail accounts or add custom words to the list manually to ensure the AI recognizes specialized terminology correctly according to TechCrunch.
Advanced Text Transformation and Analysis
Once a transcription is completed, Google AI Edge Eloquent provides several AI-powered tools to transform the raw text into a specific format. Under the transcript, users can select from several options to rewrite or summarize their speech:
- Polish: Transforms disorganized speech into flawless text.
- Key points: Extracts the most important information into a summary.
- Formal: Adjusts the tone of the text for professional environments.
- Short/Long: Adjusts the length of the output based on the user’s needs as reported by Mashable.
The app also maintains a comprehensive history of transcription sessions. This history tab allows users to search through previous recordings and view specific metrics, such as the total number of words spoken, words-per-minute speed, and words dictated in the most recent session according to TechCrunch.
Comparison of Processing Modes
| Feature | Local Mode (Default) | Cloud Mode (Optional) |
|---|---|---|
| Model Used | Gemma (On-device) | Gemini (Cloud-based) |
| Internet Required | No (Offline) | Yes |
| Privacy | Data stays on device | Processed via cloud |
| Primary Goal | Responsive dictation | Advanced text cleanup |
Market Positioning and Future Updates
The release of Google AI Edge Eloquent places Google in direct competition with a growing field of AI-powered dictation tools, including Willow, SuperWhisper, and Wispr Flow according to TechCrunch. By offering a free, unlimited, and offline-capable tool, Google is leveraging its open-weight Gemma models to attract users who prioritize privacy and cost-efficiency.
While the app is currently limited to the iOS platform, Google has indicated that further integration is planned. According to the App Store listing and updated reports, a seamless iOS keyboard integration is “coming soon,” which would allow users to utilize these AI dictation capabilities across other applications on their iPhones via the App Store.
The app is currently 67.1 MB in size and is rated for users 18 and older via the App Store.
The next confirmed development for the app is the release of the iOS keyboard integration, which Google has listed as an upcoming feature. We will continue to monitor the App Store for the official rollout of this update.
Do you use AI for dictation? Let us know your thoughts in the comments below or share this article with your colleagues.
Worth a look