A project's Translation Settings tab controls how that project auto-translates, confirms, and locks segments — which sources feed auto-translation, when a segment gets confirmed or locked automatically, and which Sous Chef Agent shapes the project's AI output.
​

💡 Who is this for? This guide is for Project Managers and Account Admins who need to configure how a project auto-translates, confirms, and locks segments within the wxrks platform.

When a project is created, its Translation Settings start as a copy of the parent Organizational Unit's Translation Settings template. You can still change them on the project itself after that — but changes only affect segments translated from that point forward; they aren't applied retroactively to segments already auto-translated, confirmed, or locked under the old settings. Review these settings before approving a project, not after.
​

You'll find this tab on the project's page, alongside Overview, Costs, and the rest — see Project Overview for a map of every tab. Two groups of related settings — Unlock options and Mismatch Warnings — don't actually live on this tab; see the last section below for where to find and change those.

Translate empty segments only: When on, any auto-translation source (Machine Translation, Translation Memory, or Context Sensitive) only fills in segments that are still empty — it won't overwrite anything a linguist already typed or that a previous auto-translation pass already populated.
​

Translate not confirmed segments only: When on, auto-translation skips segments that have already been confirmed, even if they're empty or would otherwise match. Turn this off if you deliberately want auto-translation to re-run and refill confirmed segments — for example, after fixing a Translation Memory or reconfiguring the auto-translation sources below.
​

Complete task once all segments are confirmed: When on, a task is delivered automatically the moment auto-translation confirms 100% of its segments. This only fires from auto-translation — manually confirming the last segment yourself does not trigger automatic delivery.
​

Turning this on reveals a second toggle, Complete all subsequent tasks upon task delivery: lets every other task that also reached 100% completion during the same auto-translation run get delivered too, instead of only the task the run started on.
​

Show Machine Translation suggestion in Editor: Shows a machine-translation suggestion in the Editor's Suggestions tab for every segment — even ones that weren't auto-translated by MT. Each time you turn it on, wxrks shows a disclaimer about sending content to the MT provider that you need to accept for the setting to take effect; turning it off doesn't prompt anything.
​

Penalty for multiple 100% match: A slider from 0% to 100%. When a segment has more than one 100%-match Translation Memory entry, wxrks reduces their match score by this percentage instead of auto-filling from an ambiguous match — for example, a 10% penalty turns a 100% match into a 90% match whenever more than one TM entry ties for that segment. Leave it at 0% to disable the penalty.

These three toggles are the actual sources of auto-translation (also called Pre-Translation) for the project. At least one has to be on for auto-translation to run at all.
​

Context Sensitive: Runs the project's translation through wxrks' AI, which already blends Translation Memory, Machine Translation, and Glossary matches into one context-aware suggestion — shaped by the project's Sous Chef Agent (see below). Because Context Sensitive already draws on TM and MT internally, you don't need to also enable them just so their content gets considered — turn on Machine Translation and/or Translation Memory in addition only if you also want plain TM/MT matches applied on their own where AI doesn't fully resolve a segment. Context Sensitive can take noticeably longer to run than TM or MT alone, especially on larger files.
​

If your account has multi-step Context Sensitive translation enabled, turning this toggle on reveals a Context Sensitive Priority panel where you can enable/disable and reorder the individual context providers (Machine Translation, Sous Chef Instructions, Translation Memory, Glossary) that feed into the AI's output, with a warning if you move Machine Translation off the bottom of that order. This is an advanced, account-level-gated option — most projects won't see it, and it's covered in depth in Configuring Sous Chef Agents in wxrks.
​

Machine Translation: Auto-translates segments using the project's configured Machine Translation provider.
​

Translation Memory: Auto-translates segments from the project's Translation Memory. Turning it on reveals a threshold slider — only segments matching at or above that percentage get auto-populated from the TM. Tag-correction options for TM suggestions live with each Translation Memory's own settings on the project's Context tab — see All about Translation Memories.

Prioritize TM matches over Context Sensitive translation: Only selectable when both Context Sensitive and Translation Memory are on above — it's greyed out otherwise, as in the screenshot, where Context Sensitive is on but Translation Memory is off. When enabled, it also reveals a Minimum TM match threshold slider (1%–101%): a segment's Translation Memory match has to meet that percentage before it wins out over the Context Sensitive suggestion.
​

The four toggles below follow the same rule — each is disabled until a Translation Memory match is actually able to win a segment, which means either Context Sensitive is off, or it's on and Prioritize TM matches over Context Sensitive translation is also on:
​

Auto confirm pre-translated segments: Automatically confirms any segment that auto-translation populated during project setup, without a linguist needing to review it first.
​

Confirm segments with 100% match / Confirm segments with 101% match: Automatically confirms a segment the moment it's auto-filled from a 100% (or 101%, i.e. matching source context too) Translation Memory match.
​

Lock segments with 100% match / Lock segments with 101% match: Automatically locks a segment auto-filled from a 100% (or 101%) match, preventing further edits unless it's explicitly unlocked.

Alongside the toggles above, this tab has a Sous Chef Agent selector — the AI agent (tone, audience, formality, and custom instructions) that shapes this project's Context Sensitive translation and auto-translation output. By default a project inherits its Organizational Unit's default agent (or the account's, if the Org Unit doesn't have one), but you can assign a project-specific agent here instead.
​

Configuring agents themselves — creating one, cloning an inherited one to customize it, and how account/Org Unit/project agents cascade — is its own topic: see Sous Chef Agents for how the hierarchy works, and Configuring Sous Chef Agents in wxrks for step-by-step setup at each level, including this project-level selector.

If you're looking for Unlock options (unlocking segments with numbers, special characters, or both from auto-locking) or Mismatch Warnings (ignoring Glossary or Tag mismatch warnings in the Editor) — those aren't on a project's Translation Settings tab. They're configured once, in the Organizational Unit's own Translation Settings template, and are only applied to a project at the moment it's created; changing them afterward has no effect on projects that already exist.
​

See All about Organizational Units for where to set them.

Project Overview — a map of every tab on a project's page.
​All about Organizational Units — where the Translation Settings template a new project inherits from is configured.
​All about Translation Memories — configuring the Translation Memories a project draws on.
​Sous Chef Agents — how agent settings cascade across account, Org Unit, and project.
​Configuring Sous Chef Agents in wxrks — step-by-step agent setup at every level.