Google has quietly released a new AI-powered app called Google AI Edge Eloquent, marking a significant move into “Edge AI”—where processing happens entirely on your device rather than in the cloud. Designed primarily for productivity, the app converts speech to text in real-time and functions without an internet connection.
Core Features & Capabilities
-
Offline First: After an initial download of the Gemma-based speech recognition models, the app works entirely without Wi-Fi or data.
-
Intelligent “Cleanup”: Unlike standard dictation, Eloquent automatically removes filler words like “um,” “ah,” and “uh,” as well as mid-sentence self-corrections.
-
Text Transformation: Once transcribed, you can instantly rewrite the text into various formats:
-
Key Points: Bulleted summaries of your speech.
-
Formal: Polished prose suitable for emails or documents.
-
Short/Long: Compressed or expanded versions of your notes.
-
-
Privacy Centric: By default, all machine learning stays on your phone. However, there is an optional Cloud Mode that uses Gemini for even deeper text refinement if you have an active connection.
Personalization & Integration
One of the app’s standout features is its ability to learn your specific “language”:
-
Custom Dictionary: You can add specific jargon, technical terms, or unique names to ensure the AI recognizes them correctly.
-
Gmail Sync: Optionally, the app can securely scan your Gmail data on-device to learn the names and keywords you frequently use, making its transcriptions more accurate over time.
-
Metrics & History: It tracks your “words-per-minute” speed and total word counts, saving your session history for easy searching later.
Availability & Competition
The app is currently available for iOS (requiring iOS 16.0 or later) and macOS (for M1 chips or newer). While there is no official release date for Android yet, Google’s App Store listing suggests it is evaluating other platforms, where it could eventually serve as a system-wide default keyboard.
Market Context: This release puts Google in direct competition with emerging AI-first dictation tools like Wispr Flow, SuperWhisper, and Willow, offering a free, privacy-focused alternative with no usage limits.
Tip: To get started, download the app from the App Store and ensure you trigger the Gemma model download while on Wi-Fi to enable the offline features.
