Limitations and roadmap

Current limitations

No import button. Recordings you already have can only be imported through the HTTP API; see Import recordings you already have.
v0.1.0's UI is in Chinese only. Since then the web UI comes in Chinese and English (switch it on the 「我的」 (Me) page) and the app follows the system language since 1.5.1.
You build the iOS app yourself. The app isn't on the App Store; an app installed with a free Apple ID has to be installed again after 7 days.
The LLM and the default ASR run in the cloud. Running fully local needs ENGINE=local (worse than Qwen on far-field audio, see ASR benchmark) plus an LLM server you run yourself.
v0.1.0 can't undo marks or merges. A to-do's 「完成 / 不对」 (done / wrong) marks and a speaker merge can be undone on main (tap 撤销 (Undo) in the notice), but not yet in v0.1.0.
Short conversations get no summary. A conversation with under 60 seconds of speech doesn't go to the LLM; its first sentence becomes the title.

Import: import recording files directly from the app or the CLI.
Desktop split view: the list on the left, the details on the right on wide screens.
People cards: add what you talked about lately to the existing people cards.
Player polish: keep improving the conversation page's bottom player and follow-along highlight.

The roadmap changes; the latest discussion is in GitHub issues.