Tencent's new AI voice assistant, Chatterfly, has quietly started internal testing, with media reporting that the product is the previously rumored "Baoyuan Input Method." Currently, Chatterfly supports macOS and Windows desktop versions, but requires an internal test code to use. The iOS and Android versions show "Coming Soon."

image.png

Not just voice-to-text, it can also polish content according to scenarios

Different from traditional input methods that mainly convert voice into text, Chatterfly supports AI typing across all scenarios, with strong intent understanding capabilities and the ability to continuously learn user input habits. It can understand user intent and current scenarios based on context, automatically organizing and polishing content — users don't need to pre-organize their language; after speaking, it can automatically sort out content based on context, turning spoken words into clearer and more appropriate written text, achieving "speak and write directly."

Currently, Chatterfly has six built-in scenario Skills, including meeting minutes, work reports, project progress, marketing copy, email polishing, and Vibe Coding prompt optimization. After activating the corresponding Skill, it can perform more in-depth content organization and polishing for different usage scenarios. In addition, it supports text creation, instant translation, and information summarization functions, allowing users to issue commands by simply speaking their needs with a shortcut key.

Chatterfly can be called within multiple applications and scenarios such as WeChat and QQ, without needing to frequently switch between apps, and it is compatible with weak networks, noisy environments, and soft voice conditions.

According to media reports, the product uses Tencent's self-developed HuanYuan multimodal technology at its core. Voice transcription records and audio files are stored locally on the user's computer and not uploaded to the cloud.