Intelligent Virtual Assistants (IVAs) have become an indispensable part of modern customer service, enabling businesses to interact with their customers seamlessly and personally. However, developing IVAs that can engage in human-like conversations requires significant time and effort, often requiring teams of developers and domain experts to work together to create and maintain the IVA’s conversational capabilities.
The latest release of the Kore.ai XO Platform v10.1, promises to change this paradigm by leveraging the power of generative language models to enable businesses to create intelligent conversational experiences 5x faster with 3x fewer operational efforts. Let us take a closer look at the platform’s new features and how they can benefit enterprises while building virtual assistants.
Smart Co-pilot for IVA Development
Building a virtual assistant from scratch can be daunting, requiring a deep understanding of the domain, user behavior, and business processes. The Smart Co-pilot feature of the XO Platform aims to ease this process by providing smart use case suggestions that help developers get a head start in the development process. By providing the domain and a brief description of the IVA, developers can discover all the relevant use cases that the IVA must handle, significantly reducing the time and effort required to identify the critical functionalities.
Conversation Preview & Auto-Dialog Generation
The Conversation Designer in the XO Platform now comes equipped with advanced LLM and generative AI capabilities that can automatically generate a conversation design based on the use case descriptions. Developers can then review, make necessary edits, or regenerate the conversation design until satisfied. The XO Platform then automatically generates executable dialog tasks, including prompts to collect user input, API call definitions, business logic, and transition flows based on the use case description, simplifying the entire development process.
{% video_player “embed_player” overrideable=False, type=’hsvideo2′, hide_playlist=True, viral_sharing=False, embed_button=False, autoplay=False, hidden_controls=False, loop=False, muted=False, full_width=False, width=’1366′, height=’760′, player_id=’110954893997′, style=” %}
Training Data Suggestions
Creating effective training data is a critical task that requires continuous effort to ensure that the IVA can handle all possible variations of user input accurately. The Kore.ai XO Platform now integrates with language models to provide training data suggestions that cover all possible variations that end-users may use, significantly reducing the time and effort required to generate training data. This feature also provides a wide range of training utterances to improve the intent detection rate.
{% video_player “embed_player” overrideable=False, type=’hsvideo2′, hide_playlist=True, viral_sharing=False, embed_button=False, autoplay=False, hidden_controls=False, loop=False, muted=False, full_width=False, width=’1366′, height=’768′, player_id=’111330626852′, style=” %}
NLU Test Data Suggestions
Evaluating the NLU performance of an IVA requires a comprehensive test suite that can include unseen phrases, entity checks, spelling mistakes, and negative cases. The new Generative AI capabilities of the Kore.ai XO Platform analyze the IVA scope and generate test data for various scenarios, helping developers achieve higher coverage. The platform also suggests possible user responses to simulate conversations for complex scenarios, continuously updating the suggestions as developers continue the conversations.
Learn More:
Explore Kore.ai Generative Language Models
AI-Assisted and Adaptive Dialog Node
Enabling IVAs to handle complex tasks and provide engaging experiences to users requires adaptive dialog nodes that can orchestrate dynamic conversations. The AI-Assisted and Adaptive Dialog Node feature of the Kore.ai XO Platform empowers developers to create conversational agents that use context information to provide personalized experiences to users.
{% video_player “embed_player” overrideable=False, type=’hsvideo2′, hide_playlist=True, viral_sharing=False, embed_button=False, autoplay=False, hidden_controls=False, loop=False, muted=False, full_width=False, width=’1920′, height=’1080′, player_id=’111330806283′, style=” %}
LLM-Based Knowledge Graph
Creating and maintaining a knowledge base for a virtual assistant can be a challenging task, requiring significant manual effort to create intents and define entities. The Kore.ai XO Platform offers an innovative intent-less model that uses the power of large language models and generative AI technologies to eliminate manual intent creation and answer user questions contextually.
This feature can take content from diverse sources like webpages, PDF files, and CMS pages and efficiently generate contextually relevant answers without the need for manual FAQ creation. The platform also offers a knowledge graph that uses advanced language embeddings to identify the correct answers to questions without the need for complex ontologies and extensive training.
The Kore.ai XO Platform v10.1 Release includes several enhancements support for Azure Open AI Integration, Twilio Verify and Nice In Contact Integration and support for the Danish language.
In summary, the latest release of the Kore.ai XO Platform offers innovative tools that leverage the generative language models to simplify development efforts and dynamically build natural conversations.
Check out more about the upcoming release here.
