Listen on the go!

Technologies like AI and automation are the backbone of the fourth industrial revolution. They aim to augment our professional lives not just by improving productivity but also by providing a better user experience in the man-machine interaction.

Enterprises have realized the importance of conversational ai and that it will soon become a part of a consumer’s life. The growth of the conversational AI market is promising and hence many players are foraying into this space. Gartner estimates about 2,400+ vendors in the conversational AI space. This has made the market overcrowded. In such a scenario it becomes very important for organizations to select the appropriate solutions that are in line with their vision and enterprise requirements. has been leading the conversational AI space for many years now. Renowned analysts such as Gartner, Forrester, ISG, Everest Group, DMG Consulting, and many others have recognized platform as the world’s top virtual assistant platforms. With every release, has delighted its customers with new features and enhancement and we continue the legacy of providing enriched user experience in this version as well Version 8.0 of Virtual Assistant platform comes with a package of new features, enhancements, and a few bug fixes. The blog is intended for business users, tech enthusiasts, or bot developers who wish to learn the advancements in the bot platform and how it can uplift the current user- bot interactions.

Here is the roster of key features and enhancements for this release:-

Containment Metrics

One of the key concerns of organizations while deploying the VAs is to justify the return on investment (ROI). They remain skeptical about the performance of the bots and more importantly they are keen to gauge the effectiveness of conversational AI projects. This issue has been further taken care of in this release. platform now provides an out-of-the-box feature called Containment Metrics for an in-depth analysis of customer engagement.

Containment metrics can be used to get deeper insights into the bot’s performance and the level of engagement of bots with the customers.

This feature segregates the bot-user conversation into three categories:- 
  • Self Service Conversation  – The user successfully gets the queries answered by the bot 
  • Agent Transfer Conversation – The conversation is transferred to a live agent 
  • Drop-offs – The conversation is not completed and ended abruptly (due to technical error/user did not respond/bot could not identify the intent)

The containment metrics allows the users to get a detailed break-up of the conversations over a specific time period. These metrics can be used to get insights about the trends of self-service, drop-offs, and agent transfer conversations. 

Containment metrics also provide detail on the level of bot engagement. As a part of the engagement analysis dashboard, we can see the percentage of the total conversation with respect to the duration of the conversation, no. of messages exchanged, and also the no. of tasks completed. Moreover, the parameters can be further customized (using the horizontal slider) for deeper analysis.

Containment Metrcis GIF (1)

Containment metrics feature of Virtual Assistant Platform Ver 8.0

Containment metrics come in handy for quick analysis of the successful implementation of the virtual assistants. This feature will be of immense use to the business users who wish to showcase the progress of a Conversational AI project and analyze the improvements in the VA’s performance.

Natural Language Processing 

The new version enables users to have better control while customizing the Machine Learning (ML) models. At the time of training the virtual assistants, the users can configure the hyperparameters to customize the Natural Language Processing (NLP) Models. They have the option to choose the type of neural networks (CNN/ Word Embeddings/ LSTM etc.) depending upon the bot tasks.

Thus choosing the right type of algorithm specific to the use cases will further improve the accuracy of the bots and enable them to tailor their response according to the conversation.

The developers are also given an option to choose between the n-gram approach or skip-gram approach for intent detection. This flexibility will enable them to improve the training of the bots even with a limited training data set. 

In addition to this, the platform now supports K-fold testing for the evaluation of the model on the training data.

NLP image

Advanced NLP Configurations of Virtual Assistant Platform Ver 8.0 

With these improvements, the virtual assistants developed on Virtual Assistant Platform 8.0 are made more natural with suited responses as per the context of the conversation.


Storyboard is the key component of the conversation designer that is used to capture complex business requirements as a series of conversations between a bot and a user. These dialogs are captured in the form of scenes within the storyboard. 

The latest release of the VA platform allows the bot designers to link the scenes. They are given an option to view the scenes and link them to the current scenes. By linking the appropriate scenes, they don’t have to create the same series of conversations again. They can simply link it to the relevant scene to create the overall story. Within the storyboard, you can now add tags to the notes, and color code them for easy classification of the notes and customize the background image of the preview page.

Conversation Designer with features of release 8.0

These changes thus foster the collaboration of various stakeholders and enhance the look and feel of the design of the virtual assistant.

Universal Bot

The Universal Bot is made more efficient with reduced development and training efforts. The new version of the platform allows users to mark the linked bots of the Universal Bot as “inclusive bots”. The bots which are marked as inclusive shall always qualify to get the utterance from the Universal Bot. Thus if the users require some specific linked bots to always process the requests received by Universal bots, it can be done by marking those bots as “inclusive bots’. These inclusive bots need not be trained with sample utterances to participate in the bot scoping process.

Universal Bot-1

Version 8.0 allows the developers to choose the linked bots as fallback bots / inclusive bots

Besides this, significant improvements are made in the fallback bots within the linked bots. With these additions, Universal Bot 2.0 is more robust in addressing the diverse enterprises’ needs.

Knowledge Extraction

Creating bots from long semi-structured documents is no more a tedious task. The new version of the platform comes with a built-in annotation tool that can extract data from PDF documents of various formats. The users have to just annotate a few portions of the documents. Using the ML model the annotation tool can intelligently learn and ingest the relevant information from the semi-structured documents. With this smart knowledge extraction process in place, enterprises can plan to develop bots with a large no. of FAQs with minimal efforts. 

Annotation tool 1 (1)

Annotation Tool for FAQ extraction process for semi-structured documents in version 8.0 

This feature addition enables the developers to build Q&A type of bots using unorganized PDF documents.

User Analytics

Many times it becomes difficult for business analysts to devise the right inference from the conversations. This might be because they are not aware of the previous conversations or they do not have full information about the customers. They thus draw half-baked conclusions because of the limited information. The advanced user analytics feature now allows the platform users to view all the information about a conversation  in a single interface Information such as conversation channels used, custom meta tags, intent detection rate, goal completion rate, the flow of the conversation, and many more can be analyzed now in the chat history.

Usr Analytics (1)

Advanced user analytics of Platform Version 8.0 

The feature thus enables analysts and business users to perform a better analysis of the conversation and decipher the root cause of drop-offs/agent transfers performed during the conversation. 

Other Major Additions and Enhancements 

Skype for Business on-premises and  AudioCodes channels have been added to the roll of channels that the platform currently supports. 

Additionally, significant improvements are made in Russian, Kazak, and Japanese languages. 

This release is a giant leap towards building impeccable virtual assistants. The designing, training, and testing of the VAs have been further simplified and they are made more capable to emulate human-like conversations. This release aims to improve the performance of the virtual assistants, robust analysis of user engagement, and simplify the virtual assistant building process. 

The features mentioned above are some of the highlights of this release. We have a host of other additions and enhancements for this release. Please refer to product documentation to learn more.

You can also talk to our experts for further other information related to Virtual Assistant Platform v8.0 release.

Request a Demo

Source link

Leave a Reply

Your email address will not be published.