Release Notes - Language IO Enhanced Chat for Salesforce
These are the Release Notes for Language IO Enhanced Chat for Salesforce.
Please see the installation instructions in Installing Language IO Enhanced Chat for Salesforce.
March 10, 2026: 1.21
New
Introducing the Toxicity Shield
Language IO is proud to announce a new feature that will help Agents deal with potentially problematic end users: the Toxicity Shield.
The Toxicity Shield feature is an automated, LLM-based moderation tool that assesses the toxicity of a message to assign it a score, then take certain actions or not, as applicable, based on this score.
The Toxicity Shield runs when the message is ready to be delivered to its recipient (this means after the translation, if a translation was performed). How it works can be broken down into two parts:
- Assessing and assigning a toxicity score to a message.
- Use this score as a trigger for certain actions:
- displaying an icon along with a chat message to indicate the "temperature" of the message
- Flagging, rewriting, or even blocking a message depending on a sliding threshold.
The Toxicity Shield runs on both inbound (end user to agent) and outbound (agent to end user) messages.
- When a message from the end user is rewritten, the agent can click on the "eye" icon to view the original message. The translation of this original message is displayed between the customer message and the rewritten message.
- When a message from the agent is blocked, this is clearly indicated in the user interface, so that they know that they need to take a follow-up action:
- If the message from the agent is only blocked, the agent gets a prompt to rewrite the message.
- If the message from the agent is blocked and rewritten, the rewritten message shows as a suggested reply, which the agent can either accept and send, edit and send, or reject and send the original.
To enable this new feature in Salesforce Enhanced Chat, see Enabling the Toxicity Shield.
Creating Toxicity Shield Reports
With the release of the Toxicity Shield, you can now generate reports in Salesforce on Toxicity scores returned by the server so that you can analyze the performance of the feature. Specifically, you can create reports on percentages (for example, identifying that 5% of messages received a medium toxicity score).
Reports can include:
- the original message
- the direction of the message (outbound or inbound)
- the suggested rewrite (if applicable)
- whether the rewrite was accepted, modified and sent, or cancelled
- the translation
- the score for each message (either category specific, or a combined score for all categories, provided by the server)
For more information, see Creating Translation Reports (Enhanced Chat).
Fixed
Translation entries not associated with a parent session
A bug was found in reports where on some instances, a number of translation entries would not appear in the Salesforce report. These entries were not associated with a parent session, when it is expected that reports link each messaging session with all its relevant translation entries.
This is now fixed.
January 30, 2026: 1.20.1
Fixed
Transcripts not visible after Salesforce update
We released a patch to fix an issue caused by a Salesforce update. This issue caused transcripts of previous Language IO translations to not be visible in the LIO component when the session ended.
Install the patch below to fix the issue:
- v1.20.1
November 11, 2025: 1.20
New
Transfer to Agent: Now with list of available agents
Agents who want to use the Transfer to Agent function can now see a list of all the available agents they can transfer to. This gives them a quick and easy way to see immediately who is available to take the transferred session.
The selected agent receives a transfer request that notifies them that their colleague wishes to transfer the Messaging Session to them, so that they can confirm the transfer.
Transfer to Agent: Transfer only to Queue now available
You can now set the Transfer To Agent feature to only allow transfers to Queues.
New "Used with the native conversation component" setting
If you configured your Messaging Session record page to load both the native Enhanced Conversation component and Language IO Messaging component during messaging sessions (for example, the Enhanced Conversation component is your default component, and agents switch to the Language I/O Messaging component when translation is needed), you must enable the Language I/O Messaging component’s "Used with the native conversation component" setting.
If you use the Language IO Messaging component in a standalone way (specifically, without loading the native Enhanced Conversation component during messaging sessions), you must disable the Language I/O Messaging component’s "Used with the native conversation component" setting.
Hiding the Quick Text Action in the Messaging component
Organizations can enable or disable the "Quick Text" option in the Messaging Component, depending on the features that they want their agents to access. When disabled, the "quick text" button is not visible in the UI.
Improved
Feature parity between the Language IO and Standard component
-
Message delivered/read indicator
The Language IO component now matched the standard component and shows indicators in the UI for "message delivered" and "message read". This allows agents to easily determine if the end user has seen the agent's message and is still engaged in the chat. -
"End user" is typing indicator
The Language IO component now matched the standard component and shows when the end user is typing their reply. This allows agents to anticipate a response from the end user, prepare to respond themselves, and avoid ending the chat prematurely. -
Message timestamps in the Language IO component now include seconds
Until now, messages in the Language IO component had timestamps that showed hours and minutes (10/7/2025, 8:34 AM). To improve agent efficiency, timestamps now match the standard component and show timestamps with seconds as well (10/7/2025, 8:34:28 AM).
Error 517 is now a warning message instead of an error message.
Customer feedback indicated that having Error 517 be an error message (this is the message that pops up when the source language and detected target language are the same) throws the agents off: they may assume something is wrong when they see the error, when in reality, they can simply ignore the error and continue with the conversation.
We changed the formatting of this message so that it now appears as a warning message and not an error.
End user quality flag persists after refresh
Now, when an end user flags a message, the notification persists after refresh, so that the agent do not lose track of the history of the end user flagging their translations.
Fixed
Right alignment on Right-To-Left Language
A bug was found where Right-To-Left languages, such as Arabic or Hebrew, did not align to the right as expected. This is now fixed, and RTL languages show the right alignment.
August 28, 2025: 1.19
New
Using a field on the Messaging Session object to determine the language of an end user
Organizations can now configure the Language IO Messaging app to use a field (standard or custom) on the Messaging Session object to determine the language of an end user, in order to avoid using auto-detection for this purpose.
This feature allows orgs to reduce the number of calls to the Language IO API, optimizing their use.
Transfer to Agent: Now available in the Language IO component
The Transfer to Agent function is now directly accessible through the Language IO component. Agents can use the Transfer to Agent () located in the button row above the entry field to select another agent to send the messaging session to.
The selected agent receives a transfer request that notifies them that their colleague wishes to transfer the Messaging Session to them, so that they can confirm the transfer.
Agents can also select a Flow or a Queue to transfer the conversation to another agent.
Quality Flag for end users
End users of Web Messaging can now flag a message sent by the agent to provide feedback regarding the translation they received.
Hiding the Quick Text Action and the Customer Inactive button
Organizations can enable or disable the "Quick Text" option in the Language IO Component, depending on the features that they want their agents to access. When disabled, the "quick text" button is not visible in the UI.
August 6, 2025: 1.18
Improved
Enhanced Link Messaging Components are now accessible from the Language IO application
Enhanced Link Messaging Components are now accessible from the Language IO application to match the native Salesforce component. The Language IO application now shows a Messaging component button that allows agents to select and insert Enhanced Links that are available in the org. This new button is located between the quick text button and the flag raise button.
This way, the page can load any available enhanced links without requiring the agent to return to the native Salesforce interface.
Older versions
Submit a support request if you need to review older release notes for this application.