Qatar Release Notes

This document provides an overview of the new features and enhancements that are being introduced in the Qatar Release. Release Date: April 5, 2024

New Features

Hyperflows

Hyperflows are Aisera's new AI Automation capability. Like AI Workflows, Hyperflows automate complex end-to-end flows, integrating human and machine tasks. The key features are:

  • Fulfillment of user requests dynamically - an LLM sequences Aisera actions and API calls to external systems.

  • Capability to provide automation use cases via a prompt and generate a Hyperflow in minutes.

Business Metrics Dashboard

We introduced a new Business Metrics dashboard to demonstrate the following ROI-based KPIs for an Aisera App, including the ability to configure them:

  • User Productivity Savings

  • Agent impact

  • Avoided Cost

  • Resolution Count Metrics

Enhancements

Pre-Built Analytics

Pre-Built Analytics have been enhanced to improve accuracy and deliver more relevant visualization.

Key Feature Enhancements include:

  • Inclusion of Unhandled requests within Relevant Requests

  • Tooltips are updated to include Unhandled Requests

  • For Irrelevant requests, the Resolution field has been updated on the Request Details Page

  • Feedback has been removed from Assisted Requests

  • All Assisted requests, where users abandon the conversation before a ticket is created, are marked as Assisted and not Abandoned.

  • For any Request that encounters an error, the Resolution is marked as Unresolved by default.

  • Resolution and Resolution Type fields on the Request Details page are marked as Read Only fields.

  • Enhancements are made to improve Computational Performance and Accuracy.

Custom Analytics

Support User Custom field in Custom Analytics

  • Any custom field ingested as part of User Learning can now be used in Custom Analytics to create different visualizations.

Auto-Categorization

The Following enhancements have been made to the Auto-Categorization feature:

  • Enhanced Model & Performance:

    • 5 new tags have been introduced - Flow Execution Error, Not Understood, Incorrect Answer- Update Annotation, Update Ontology & Update Annotation & Ontology

    • We tag the requests that serve fulfillments from RAG & Public KB fulfillment sources.

    • We tag all types of Requests by default (Resolved, Casual, Assisted, Abandoned, etc.), including Unresolved requests

    • Auto-Categorization functionality is now adapted to Intentless Architecture

    • We have the ability to run Auto-Categorization job for specific request IDs

  • Enhanced Unresolved Analytics by adding following visualizations to highlight:

    • Low performing KBs and Flows

    • Intents with KB Gaps

    • Top Recommended Actions

Channels

Support RTL(Right-To-Left) Languages in Webchat (Hebrew only)

  • Aisera Webchat now has the capability to support RTL languages. The first RTL Language included is Hebrew. Users conversing in Hebrew will now experience a more natural flow of conversation, ensuring a smoother and more intuitive interaction with the chatbot.

Aisera Assist

With the Qatar release, Aisera Assist has been enhanced with the following features.

  • Upon tenant configuration, Case Summarization and Aisera Answer will now take any changes to the ticket (including comments, title, description, and fields) into account in real-time, delivering agents with the most up-to-date information.

  • Aisera has implemented optimizations resulting in significantly reduced latency during the loading of widget components.

  • Ticket Lens now offers enhanced functionality by accepting Ticket ID as input for both single ticket and batch runs. By incorporating Ticket ID, Ticket Lens comprehensively captures all facets of the ticket, including title, description, comments, and configured fields, and generates Case Summaries and Aisera Answers as part of the results.

Aisera LLM Gateway

  • Aisera now supports foundational language models available through Amazon Bedrock. We also support more Chat Completion Style APIs across various foundational language models.

Neural+

With Qatar Release, we have introduced the following enhancements to Neural+ Fulfillment engine.

  • Answers generated by Neural+ will now support URLs and hyperlinks in the source document.

  • Enhancements are made to Neural+ based answer generation to reduce latency and inconsistencies.

AI Workflows

AI Workflow capabilities have been enhanced in the Qatar release. Key features are:

Character Limit support in Workflow User Input Node

  • In AI Workflow User Input node, users can now set a character limit, which is shown to the end-user within the chat experience.

Additional Actions and improved logging capabilities

  • New Action to generate and store ticket summaries.

  • New Action for Aisera Answer generation.

  • New Action to extract intents.

  • New Action to trigger paused flows using callbacks.

  • Remote REST call execution capabilities are expanded.

  • Ticket Action nodes have Test Capability

  • Improved error reporting features for Ticket Actions

  • Improved error logging and monitoring

Connectors and Integrations

  • ServiceNow connector has been enhanced.

    • Support Interaction Learning

    • Problem learning - Close Notes field is available

  • Sailpoint connector has been enhanced.

    • Support User learning

  • Self-serve option to un-retire knowledge documents from the data source page

Voice Experience

Voice Experience Enhancements in Qatar cover:

  • Support for a fully redundant Voice Gateway orchestration.

  • Availability of Voice Activity Detection to reduce latency and detect silence.

  • Caller information is available in the Aisera Platform via Workflow Session Variables.

  • A Library of Sound files is available for playing during hold, waiting for Virtual Agent actions.

  • Introduced AI Workflow Actions for Call transfer, Call termination, and Playing sound.

  • Backend performance and connectivity scalability between Voice Gateway and Aisera platform

Other Updates in Qatar

  • Universal Bot has been enhanced with respect to Language Handling.

  • We now provide Translation Feedback by the end-user on webchat.

  • Introduced the capability to NOT SHOW feedback at the end of the Campaign.

  • The Request Details page shows whether a Service Catalog Item or a Knowledge Article has been clicked through.

  • SMS Channel now supports Images.

  • Introduced New APIs to support User Account Management.

  • Universal Bot supports SMS channels along with typical digital chat channels.

  • Improvements are made to Similarity-based Ticket Clustering Models.

Last updated