All projects

Munhae Studio

A local-first Classical Chinese translation workspace

A local-first desktop workspace for Classical Chinese translation, research, and AI-assisted review.

Role
Solo developer
Platform
Desktop application
Status
Beta
Stack
Electron, TypeScript, React, AI CLI integrations
The English product introduction, with source text and translation shown together.

Keep the translator close to the text.

Munhae Studio is a desktop workspace for people translating Classical Chinese. It brings local projects, terminology, side-by-side review, and AI assistance into the same working environment.

I develop it solo. It builds on an earlier batch-translation tool I made in 2025, and shifts the focus from producing a first draft to helping someone research, review, and refine it.

Understand the work before changing it.

Getting an initial translation is one step in a much longer process. A translator still needs to consult the source, maintain terminology, compare choices, and decide what to keep.

The workspace needs to preserve that context while making AI assistance available as part of the process. A generated draft still needs a person’s judgement.

The choices behind the product.

01

Organise the work in local projects

The desktop application keeps translation work in local projects. Source material, terminology, and review belong to a continuing piece of work rather than a disposable chat session.

02

Put source and translation side by side

The review interface keeps the original text close to the draft. This helps the translator inspect wording and revise a passage without losing the source context.

03

Connect the user’s own AI tools

AI assistance integrates with the user’s Claude, Codex, or Antigravity CLI. The application provides the translation workflow around those tools.

Learning through the work.

Munhae Studio is a beta product. Its earlier batch-translation tool shortened one translator’s first-draft work from weeks to hours; that experience motivated a fuller research and review workspace.

The design challenge is to make assistance useful while keeping the translator’s decisions visible and editable. I continue to develop the product around that balance.