Read automatic email activity and case chats
Open a message after selecting its mailbox. Three indicators below the sender show sorting, priority, and automatic-reply activity. Feature read permissions control the available information.
Only an actively executing phase has a moving colored border. Queued, disabled, complete, failed, and waiting-for-decision states are static. Their tooltips work with the keyboard. Reduced motion also makes active states static. Progress comes from work state, not elapsed-time estimates.
The mail response projects existing SortingWork, SortingArrival, and AutomaticReplyCase records. Temporary phase metadata distinguishes priority assessment from sorting. Attempt leases reject late writes by older workers. Final results replace phase metadata. Batched page queries select the latest sorting work without separate polling for each row.
Open automatic-reply activity
Open Sorting → Auto reply to see the last 100 cases. The table shows source subject and sender, timestamps, matching and selected rules, decision reason, and status. Rule names come from the stored case snapshot. Later renaming does not change history. A deleted source message has an explicit indication.
Open conversation opens the case chat with its account context. Settings contains configuration and links to this activity view. Automatic replies use their own read/manage/send permissions. They do not inherit sorting permissions. See configuration and permissions.
Drafts and chat continuation
Case work uses the durable decision process and canonical draft handling. Chat progress must distinguish recorded input from completed draft work. A recorded message alone does not prove that a draft changed.
A historical draft attachment opens the current canonical draft. Opening an older proposal does not apply that text again. An already open editor retains unsaved changes. Revision conflicts appear as errors.
After reload, a locally restored editor can have an unknown save baseline. Autosave and chat submission must not automatically write that older content back. The attachment opens the canonical version for reading. Loading the newest version into the editor requires an explicit action.
Mention a person
The chat @ picker uses the composer's contact search and chips. It attaches structured contact/email references for at most 20 people. The server resolves names from currently accessible mailbox contacts and checks access again before later model calls.
A mention sends no notification or email and grants no new agent permission.
Manual priority and label selection
The email context menu's Priority submenu shows the mailbox's MailGen priorities, colors, ranks, and a clear option. This is a local priority, separate from Outlook importance and categories. Manual selection preserves scam assessment. A later requested sorting run can reassess priority. The server scopes every message and definition to the mailbox, and realtime updates use that account.
Single and bulk label pickers use a scrollable Command list. Long lists remain scrollable within the picker. Toolbar priority sorting, select-all, and timeline collapse controls use shared tooltips. Checkbox labels and keyboard behavior remain available.