Account and plans
- Workspaces and roles: who sees what.
- Plans and allowances: the five plans, AI words, packs and trials.
- Billing: invoices, the billing portal, offer codes.
- Your preferences: the choices that follow your account.
Settings
Settings is the workspace's own: what its admin may change for this workspace, and nothing platform-wide. It holds what the workspace decides: members, pass marks, which AI providers may read your text, branding. What one client needs goes in that client's profile. Some values are set by the platform and shown read-only, with a small lock.
Each setting is one row: its name, the control, then its state (such as Your choice, Saved or Set for this workspace). An i beside the name explains the few settings whose effect the name does not say, such as the second opinion or the pass marks. Its tabs are Plan and usage, Members, Workspace, Branding, Security and data, Quality, AI review, Connections and Developer, in that order, under the group names Workspace, Quality and Connections.
| Tab | What it holds |
|---|---|
| Plan and usage | This month's AI-reviewed words against the allowance, the words by source, the plan card and the Billing card. |
| Members | Who is in the workspace and their roles. |
| Workspace | The defaults, the default fallback profile and the Account defaults card. The fallback profile is what a job runs on when no profile matches it; with none picked, a file in a language with a default profile runs on that one and anything else on generic. A picked fallback comes before the language defaults. |
| Quality | The workspace defaults for the QA score, every check's severity, the LQA template and a new bake-off. See Quality defaults. |
| AI review | The AI switch, which providers may read your text, the primary model, the second opinion and its model, the daily AI runs per person and the AI check of rule findings. |
| Connections | What the workspace plugs into: the Dropbox accounts and the folders they watch, Smartling, Crowdin and the assistants connected over MCP. |
| Developer | The API keys, the webhooks and the API documentation. |
| Branding | The white label. |
| Security and data | The roles table, the access posture, your data, the retention setting, the export and, last, the Danger zone with Delete workspace. |
Your own notifications and the browser extension connection are under your name in the left menu, User settings.
The Integrations page in the left menu (Team and up, PM and up) is the catalogue: one tile per way a check can arrive, its status and a link to the tab above that sets it up.
Quality defaults
Settings, Quality is where a PM or an admin sets the defaults every profile in the workspace inherits. A profile can still set its own value for anything here; the profile wins for its own files. Below PM the page is view only and says who to ask.
The page has three parts. LQA and Eval show only when the workspace has that module.
What's Wrong
- QA score: the penalty per finding and the two marks. A critical finding costs the first number, a major the second and a minor the third, 0 unless you change it. A file passes at or above the pass mark, fails under the review mark and is for review in between. An open critical fails the file whatever the score. The line under the marks previews the score of a sample file of 2,000 words with the numbers as they stand. Change a number and press Save; Reset to shipped defaults puts the numbers back.
- Rule severities: every check the engine ships, in families such as Numbers and units or Tags and placeholders, each with the severity it runs at in this workspace: Critical, Major, Minor or Off, the words the report uses. Pick a value, and the row shows the shipped default beside it with a reset of its own. Save writes every change at once and counts them for you; Reset all to shipped defaults clears every pin on the workspace. A line under the header names any profile with no active check and any profile that pins values it would inherit anyway; when everything is fine, no line is shown.
LQA
- Default LQA template: the scorecard a new LQA job starts from when no profile, account or division names one. Pick a built-in template or one of the workspace's own and press Save; the bands and weights of the chosen template are shown under it. The templates themselves are edited in LQA, Templates, where a built-in one is duplicated before it can be changed.
Eval
- New bake-off defaults: the engines a new bake-off starts with, the source language and whether deep grading is on. Whoever starts a run can still change these on the form.
A link to a part of the page can be shared: #settings/quality, #settings/quality/lqa and #settings/quality/eval.