Four switches on a project decide how much AI does by itself and how hard it works.
Who this is for
Owners and admins. Everyone else sees the settings read-only or not at all.
What happens next
Settings → AI translation holds four switches.
Auto-translate: "Automatically queue translations when the source text changes." With it on, new and changed source strings are drafted without anyone pressing a button. With it off, translation happens only when someone runs it from the editor.
Quality mode: "Use the higher-quality model for every target language in this project." Better output, and it is also what unlocks confidence-gated review for the project.
Translation memory: reuse an identical translation instantly instead of paying to translate it again. When the exact same source text has been translated before, transglot reuses that translation at no cost.
Semantic translation memory: "Find near-matching human-reviewed translations to suggest to reviewers and to ground the AI. Fuzzy matches are never applied automatically." This is the important caveat: a similar match is a suggestion in the editor's context panel and a hint to the model. It is never written into your project on its own. Available on Team and above.
If it doesn't work
A switch marked Available on Team and above is disabled on lower plans.
If every switch is disabled, you do not have permission to change this project's settings.
Related
Translate with AI
Translation memory
Why was my translation flagged for review?