Workspace settings

A workspace is a tenant: its own members, chats, groups, agents, Brains and settings. Its settings live on the Admin console's Workspace page.

Who can change what

Every setting has a lowest role that may change it. Anyone at that role or above can change it; the server checks each setting on every save.

Lowest role	Settings
Facilitator	Findings during interviews, Founder answers in Topics, Founders and the company Brain
Admin	Trash retention, API keys approving findings, Facilitator access notices, Live AI costs, the two Notification emails switches
Owner	Everything else on the page, including data residency, deliberation, read receipts, names and labels, and the composer

• An Owner sees the whole page. An Admin sees it too, and the settings above their role say "You can see this setting but not change it."

• A Facilitator who is not an Admin opens the page from Admin > Workspace, or from Workspace settings in the F role badge at the bottom of the left sidebar. Their page shows only the settings their role can change, under the line "Only the settings your role can change are shown here."

• A few sections carry their own rule, named in the section below (the workspace name and tile, the brand, the link lists and Facilitator access to private chats).

• Every setting here applies to this workspace only. Where a setting has a Use platform default button, pressing it clears the workspace's own value and the workspace follows the platform's value again.

• Platform-wide settings belong to your platform operator and are not on this page.

Settings take effect when saved. Nothing needs restarting.

The sections of the Workspace page

Workspace

The workspace name and tile, as everyone in it sees them.

• Edit workspace opens the name and the workspace type: Business, Personal or Client. The type groups the workspace on the rail.
• Anyone who is a Facilitator or higher (Facilitator, Admin, Owner) can change the name and type.
• Click the tile to open the avatar editor. Every member can change the workspace image there.

Initials

The tile shows initials made from the workspace name. Where two of your workspaces would show the same initials, the tile is lengthened until each is unique.

• To choose your own, open the avatar editor from the tile and type up to 4 characters in Initials. Leave it blank to use the ones made from the name. The preview shows exactly what the rail will draw.
• Setting initials needs a Facilitator or higher.
• If two workspaces are given the same initials, the one created first keeps them and the other is lengthened.
• A brand logo or an uploaded image shows in place of the initials.

Brand

The name, logo and colours people see inside the workspace, in its emails and in what its agents call the app.

• Current shows the brand in use.
• Addresses for this brand here lists the App, Sign-in and Onboarding addresses for that brand. They are read-only. Your platform operator sets them.
• A workspace Admin or Owner picks from the brands available to the workspace and presses Save brand.

LEMA Logic attribution

A read-only line showing whether the workspace's app emails carry the platform operator's attribution. Only platform staff can change it.

Data residency and Overrides on file

Where the workspace's inference, embeddings and integration traffic may go, and the audited exceptions to that rule. See Data residency and personal data.

Deliberation

How many rounds a Council or Board run may go before it stops.

Setting	Default
Council max rounds	3
Board max rounds	5

A run stops earlier when the answers stop changing. See Deliberation.

Read receipts

Seen after (ms) is how long a message must be on screen before it counts as read. The range is 250 to 30000. The platform default is 1000.

Diagrams

Line style sets how connector lines in diagrams are drawn for everyone: Right-angle (the default), Straight or Smooth. Each member can still pick their own in Settings, and any diagram can be switched from its menu.

Text editor

Editor chooses the editor documents open in: Classic (Markdown source), the default, or Rich text. Changing it converts the workspace's documents in the background. The section then lists how many were converted and how many were skipped, with the reason for each skip.

Left nav sections

Auto-close other sections closes the other left sidebar sections when one is opened. It is on by default.

interviews_nav_section decides where Interviews show in the left nav:

• separate (the default): Interviews get their own Interviews section, above Topics. An Interview is a Topic with an interviewer agent or a checklist, or a programme week.
• in_topics: they stay in the Topics section.

Each person can override this for themselves in Settings → Navigation.

Composer

Placeholder is the hint shown in an empty message box for everyone in the workspace. Enter 1 to 120 characters and Save. The platform default is "Message, @tag, #asset, /command, or paste".

Names

The words the workspace uses for its sections. See Names and labels.

Message removal Undo

Undo window (seconds, 0 = no limit) sets how long a removed message can be brought back with Undo. The default is 0, which means there is no time limit.

Trusted and blocked link domains

Two lists: sites whose https links in agent replies open without the confirm step, and sites whose links are blocked.

• This workspace entries are yours to add, edit and remove. Enter a host such as docs.example.com or *.example.com.
• Platform defaults and Platform blocked are read-only.
• Facilitators and admins can change these lists.

Support access log

Every time platform support used support access on a member of the workspace, and the Notify users when support accesses their account switch. See The support access log.

Facilitator access to private chats

Facilitators can see and act in private chats lets Facilitators and above see and act in each member's private chats with agents, and their direct messages, on the member's behalf.

• On by default. Only a super admin can change it, and every change is audited. Use platform default clears the workspace's own value.
• Everything a Facilitator does there is recorded under their own name.
• Sealed interviews stay visible to Facilitators either way.
• When this is off and Founders write to the company Brain is also off, both sections show a warning: a Founder's findings would wait in a chat no Facilitator can read, so nobody could approve them. Turn one back on.

Facilitator access notices

Show facilitator access notices in private chats is off by default. When it is on, each private chat tells its owner that Facilitators can see it, with a banner above the message box and a line in the chat's greeting. When it is off there is no per-chat notice, and Facilitator access is covered in these guides and the privacy summary. A workspace Admin or Owner can change it.

Live AI costs

Show the live cost of each agent reply to admins is on by default. While it is on, workspace Admins and Owners see a live $ cost on each agent reply and the chat's running total. Members never see costs. A workspace Admin or Owner can change it. See Credits & spend.

Notification emails

Two switches decide what the notification emails sent from this workspace carry. Both are off by default, and a workspace Admin or Owner can change them:

• Notification emails: AI summary: Add a short AI summary to notification emails.
• Notification emails: recent messages: Add the latest lines of each linked conversation to notification emails.

Both are always off in a workspace that requires zero data retention for personal data, whatever the switch says. Each person still chooses their own emails in Settings > Notifications; see Search & notifications.

Findings during interviews

Show findings while an interview is running is off by default. A Facilitator, Admin or Owner can change it. Everyone else who can open the page sees "You can see this setting but not change it."

Founder answers in Topics

Count founder answers given in the Topic conversation is on by default. When it is on, a Founder's answers in a shared Topic conversation are checked against the Topic's checklist and count toward their own onboarding progress, as interview answers do. Off, only interviews count. A Facilitator, Admin or Owner can change it.

Founders and the company Brain

Founders write to the company Brain is on by default. The company Brain is the onboarding site's Brain, which every member can read.

• On: Founders save findings straight into the company Brain.
• Off: a Founder's findings stay proposed in the chat until a Facilitator accepts them.
• Facilitators and admins can always save into the company Brain.
• A Facilitator, Admin or Owner can change it.
• If it is off while Facilitators can see and act in private chats is also off, a warning appears here. See Founders and positions.

API keys approving findings

Let API keys approve findings from Agents is on by default. A finding from an Agent needs a person's approval before it goes into a Brain.

• On: an API key can approve it when the request says a person reviewed it and who reviewed it and how. The key's owner is recorded as the approver, and the review details show to everyone in the chat.
• Off: a person signed in must approve every finding from an Agent.
• A workspace Admin or Owner can change it.

Trash retention

Keep deleted items for (hours) sets how long a deleted item waits in this workspace's Trash before it is deleted for good: 24 to 720 hours, 72 by default. Use platform default clears the workspace's own value. A workspace Admin or Owner can change it. See Trash.

Export workspace

Download the workspace's content and files. See Workspace lifecycle.

Settings in Edit workspace

The pencil on the workspace header opens Edit workspace. Besides the name, type and Workspace image it holds:

• Data & model region: the EU-hosted models only and Personal data: ZDR only switches. See Data residency and personal data.
• Danger zone: archive the workspace or move it to the Trash. See Workspace lifecycle.

Settings without a control

Many more settings can be set per workspace than have a control on the page: routing behaviour, size limits, interview wording and others. Any agent in the workspace can look a setting up in the Handbook's settings reference and tell you what it does, its default and who can change it. Ask one, for example "what does the prompt caching setting do?".

Email notifications

Each person chooses their own notification emails in Settings > Notifications: the on/off switch, how often each kind is sent and where they go. What the emails carry is set in the Notification emails section above. Nobody is emailed about something while they are looking at the app.

Content scan

The content scan that checks what agents read from outside the workspace (fetched pages, guest messages, uploads) is core platform security. Only your platform operator can change it, for the platform or for one workspace.

Embedding a chat in another site

One chat can be shown inside a frame on another website.

• The address has one form: /embed/<workspace id>/chats/<chat id> on the app's own address. No other page of the app can be shown inside another site.
• Your platform operator keeps the list of sites allowed to show an embedded chat. Each brand's own onboarding site is always allowed. Ask your operator to add a site; a workspace admin cannot.
• The person must already be signed in to LEMA Chat in the same browser. There is no sign-in form inside the frame. Someone who is not signed in sees a notice asking them to open the app in a new tab, sign in, and reload.
• The address can restrict the frame, never widen it. Adding ?header=0 removes the title row. Adding ?hide= with a list of message-bar features switches those features off.
• Your operator decides which message-bar features an embedded chat offers at all. Whiteboards, documents, widgets and SideChats are never offered in an embedded chat.

Debug traces

The Admin console's Debug traces page shows what was sent to the models for each agent turn. See Debug traces.

Dispatch audit

The Admin console's Dispatch audit page shows, for one chat, what happened to each message: whether it was routed to an agent, which agents, what triggered it and why.

• Pick a chat and a Range (last hour, 24 hours, 7 days or 30 days).
• The verdicts are Routed, No agent, Skipped, Refused, Paused, Queued and No record.
• You need to be a Facilitator or higher, and you can only pick chats you can see.

Authoring courses

Creating courses, editing lessons, uploading a cover image and reordering sections needs the Manage Courses permission, which Facilitators, Admins and Owners hold. It is done in place from the course home with Edit course, or New course in the catalog.

A per-learner roster and a cohort summary for each course are available to its authors and Facilitators through the API.