On mobile, voice chat is the draw: hold to speak and the reply is read aloud, which is easier than typing on a commute or when your hands are busy. The web version covers writing, file reading and multi-turn questions. Agents are the other common entry: set the tone and knowledge scope once, then reuse them. For public-facing copy, its conversational style needs a human pass, and factual answers should be checked against sources.
What it does
- Web and mobile app, with conversation history carried over
- Voice input and spoken replies
- Create agents with fixed instructions and share them
- Upload documents for Q and A and summaries
- API access to the same models
Who it is for
- Ask a full question by voice when typing is awkward
- Keep a reusable setup for mock interviews, essay feedback or speaking practice
- Quick research, rewording or copy drafts on a phone
- Put chat into your own mini program or back office
How to get started
- Test its tone with a real piece of material in the app or on the web
- Turn repeated requests into an agent with a clear role, output format and no-fabrication rule
- For voice, start with a short question to check recognition, then move to longer material
