Fact Checking

• Feature: The chat input field in the ticket editor allows to add emoji to messages now.
• Fix: The chat selection checkboxes in the ticket history panel would sometimes be hard to reach for longer messages. The checkboxes are fixed to their message now and will stay in the viewport for as long as possible.
• Fix: Under rare circumstances, the closable state would not be determined for tickets associated to a single research only.
• Fix: Assigning a ticket to yourself would take you back to the ticket list, despite you're probably about to work with it, making it necessary to wait for the list to load, select the ticket, and wait for the editor to load. This has been fixed.
  Additionally, the assignment popup will no longer show the "Already assigned to this agent" hint if a ticket is not assigned yet.
• Fix: Deleted agents would appear in API responses and agent dropdown lists.
• Fix: Tickets or Research would always open on the same tab from the list, regardless of whether the CTRL key was pressed or not. This has been fixed so by holding down CTRL and clicking an item, it will be opened in a new tab instead of the current one.
  Additionally, the cursor will now properly indicate table rows as clickable.
• Fix: Searching for tickets or research and specifying multiple search terms would search for the entire search string as a coherent term instead of individual words. This has been fixed so the position of the individual words doesn't matter anymore.

• Fix: Research could not be created from the ticket history widget
• Fix: Responding to tickets with verbatim messages would not be possible from the ticket history widget
• Fix: Filter UI elements on the ticket list would not receive the loading state properly, causing them to stay enabled during data fetching and possibly resulting in an undefined state
• Fix: Sending webhooks for new researches would trigger an error

• Feature: Made single ticket page faster by loading data asynchronously.
• Feature: Improved ticket list layout to show more data.
• Feature: Added chat preview to ticket list. Click the chat message icon of the chat column to show the full ticket history in a popup.
• Feature: Added buttons to close a ticket or close it and mark it as spam directly from the ticket list. Simply click the new dot indicator menu to the right of a ticket and choose an option.
• Feature: Added configuration option for the spam label on the settings page. Here, you can choose a label that will act as your spam definition label.
• Feature: Added new chatblock section in ticket view. Here, you can easily use any MCP chatblock in the customer chat.
• Feature: Added new webhook events: message_sent, notification_sent, research_created, research_approved, research_rejected, research_outdated, research_assigned and research_review_requested are now available for subscription via the API or the settings page.
• Feature: Improved research list layout to be more clear.
• Fix: Fixed similar image answer suggestions. The list now includes an instant preview of all similar versions of the image.
• Fix: Under some circumstances, tickets would not be closable from the dispatch step.
• Fix: If an agent would write a free-form response on an unassigned ticket, it would not be assigned to them.
• Fix: Under some circumstances, chat messages currently being researched would not be selectable but trigger Ambiguous Selection. This has been fixed and all chats should be selectable, unless they belong to different researches.
• Fix: The ticket list would show bad performance with a large number of tickets
• Fix: In rare cases, the intelligent response assignment algorithm would behave rather stupid and assign unrelated researches to tickets.
• Fix: In rare cases, responding to a ticket would trigger a 404 - Not Found error.
• Fix: Page titles would not be set but instead default to "Dashboard", making it difficult to navigate browser history.
• Fix: Sometimes the link crawler would crash if a link in a chat contained invalid meta data, leading to repeated (and futile) crawling. If a link's meta data contained an invalid URL, it would sometimes end up differently in the database, leading to an impossibility to retrieve it by a URL.

If you have any suggestions or think you've spotted something we didn't mention in the change log, please let us know!