9.3.x Release Details

The Polish Language support is now enhanced to introduce the following changes:

• The Person Name entity is updated with the dictionary data to add more valid Polish names. This entity can also be customized to ignore unrecognized names.
• The City entity is updated to add many variations of Polish City and Village names.
• The Country entity is updated to support a wide variety of Polish utterances for a country name. For example, Polska (Poland) is a Polish utterance, which is now recognized as a valid country name.
• The Date entity is updated to support the standard Polish language date format that is DMY (Date, Month and Year). For example, if you enter a date such as one three – Twelve – Two Thousand Twenty Two (A mixture of ordinal and cardinal numbers), the platform identifies with a new date format.

To customize the Polish language entities, read the Entity Rules and Multilingual Virtual Assistant Behavior articles.

You can now customize the conversation flow when your virtual assistant identifies ambiguous intents. Use the new Ambiguous Intents Identified Event to trigger a dialog and define your custom business logic to handle the flow. The context contains the list of ambiguous intents and the confidence scores from the NLP engines. Learn more.

The Kore.ai XO Platform now helps identify incorrect patterns in the Knowledge Graph. A checkpoint Patterns with Invalid Syntax is introduced in the KG Inspect to identify syntax errors that occurred knowingly or unknowingly while defining patterns. Learn more.

The Platform now provides a new field containment_type in the Sessions Dataset of the Custom Dashboard to identify whether the session has resulted in Drop-off, Self-Service, or Agent Transfer. Learn more.

The Platform now allows the admin user to enable Two-Factor Authentication (2FA) at the workspace level for end users from the Admin Console. When accessing a 2FA-enabled workspace, the user is prompted to validate using the verification code sent to the registered email address. Learn more.

The Kore.ai XO Platform now allows developers to enable, disable, or update an existing or new language for a virtual assistant via public API. Learn more.

The Webhook v2 channel now allows you to change the language for the virtual assistant’s response for ongoing conversations. Also, you can close an active conversation session by sending the session_closure event in the Type parameter. Learn more.

Google Business Messaging (GBM) is supported as a channel for your virtual assistants. Learn more.

The Kore.ai XO Platform now supports the Sunshine Conversations (previously Zendesk Sunshine) Messaging Channel to launch your virtual assistants. OAuth-based authorization for channel integration and in-built live agent transfer to Zendesk via Switchboard are now supported. Learn more.

The Kore.ai XO Platform now allows the user to add and delete up to 20 phone numbers to the same virtual assistant for Infobip on the WhatsApp Messaging Channel. The Platform will automatically respond to the ‘from number’ (source number) in the incoming request. Learn more.

The Kore.ai XO Platform provides the discardAsCommand entity rule, which allows you not to consider the discard command as a valid string input in the String node and discards the task. Learn more.

The Kore.ai XO Platform introduces a delete method to remove the root-level object and key-value pairs inside the object of a BotUserSession. Learn more.

The Platform now allows developers to enable or disable the polling sequence in the Webhook v2.0 Synchronous integration mode.

By default, the Enable the Polling in Webhook V2 option is selected in the BotBuilder. The platform delivers consecutive bot responses to the incoming request when this option is enabled. The platform initiates the polling sequence if additional responses are to be delivered after executing any service, script, or webhook nodes in-between. Also, enabling the polling gives the option of mimicking the Web-SDK-like user experience.

Deselecting the option and publishing the bot help you retrieve the complete response without polling. This message delivery experience is similar to the V1 behavior that sends all the messages stacked in an array object. Learn more.

Get Analytics API of the Platform now supports the unhandled utterances metric type and can be used to query the data specific to unhandled utterances. The API request has a metric type – unhandledutterance, to identify whether the intent contains unhandled utterances or can be categorized under other types of NLP Insights. The response contains all the fields that are displayed on the Unhandled Utterances page. Learn more.

The new Health and Monitoring feature provides a goal-driven approach to improve the accuracy of the virtual assistant’s NLP model. The training data is analyzed along with the test coverage and test results of the Batch Test suites to provide insights into the performance of the NLP Model. The feature also recommends corrective actions to address issues identified in the training data. Learn more.

The training data validations now include incorrect intent patterns, short training utterances, and incorrect entity annotations. Learn more.

The Batch Testing feature now provides a visual summary of the test results along with the test coverage details. You can also review the test execution summary for every intent type, drill-down to specific test cases, view the expected and matched results, and view the detailed NLP analysis. You can also tag specific test case results that need follow-up actions and collaborate with your team to improve the performance. Learn more.

The new Overview Dashboard provides a summary of key metrics like conversations and containment, users, intent identification rate, goal completion rate, and more. You can also drill down to the individual dashboards to view complete information.  Learn more.

The new Users Dashboard provides the trend of the number of users interacting with the virtual assistant. The dashboard also provides the trend of new users of the virtual assistant and also provides cohort analysis of the Returning Users. Learn more.

The Analyze module now has a new navigation menu to easily navigate between various dashboards. Learn more.

The new Conversations History module feature helps you analyze the user interactions by summarizing the key events identified during the conversation. Intent identifications, entity or confirmation retries, task execution failures, etc., are some of the key events tracked. Along with the summary, you can also see the complete conversation transcript to make data-driven decisions. You can also tag conversations that need follow-up actions and collaborate with your team to resolve them. Learn more.

The new Unhandled Utterances section in the NLP Insights feature captures the utterances that did not result in intent or entity identification while executing entities or confirmations. These utterances help you identify any potential issues in the conversation design or new use cases for your virtual assistant. Learn more.

This update also includes the following updates to the NLP training modules

• The platform now supports out-of-the-box Bot Synonym lists for Simplified Chinese and Traditional Chinese languages. Learn more.
• The Address entity includes the full formatted address for you to easily parse specific parts of the address like city name, country name, etc. Learn more.
• The Phone Number entity is enhanced to accept additional formats. Learn more.

The Web SDK Client is upgraded to v2.0 to include a significant number of enhancements. Some of the key updates are:

• Customize Templates using Angular or ReactJS frameworks.
• Adopt the plug-in based architecture to integrate only the required features. For example, you can use the Graph Templates plug-in, Pickers plug-in, STT/TTS features distributed as plug-ins
• Supports easy deployment by embedding the script tag on your website code.

You can learn more about the Web SDK 2.0 from here.