Hoci

Higher-order communication

Tell hoci what to communicate; an LLM composes outgoing messages, summarises incoming threads, and handles everything around the words.

Above the words, not stuck in them

You stay at the level of intent; hoci handles the surface form. Everything around the words is part of the system, not your problem.

Describe, don't type

Tell the app what to say; it composes the message in your voice.

Replies in your voice

Suggested replies framed in how you actually talk, not generic AI prose.

Cross-language

Translate outgoing or incoming messages without leaving the chat.

Summaries on demand

Long threads collapse to a paragraph whenever you want one.

Native, fast

Rust + native webview. Tens of MB, not gigabytes. No Electron.

Familiar chat

Channels, DMs, presence, history — the boring parts still work.

Get hoci

Available for macOS and Linux. Windows coming later.