KAI

Kasisto’s digital experience platform, KAI, powers “Best in Class” – financially literate and Intelligent Digital Banking Assistants that deliver significant value to financial institutions of all sizes in retail, institutional and wealth management.

The KAI (Kasisto Artificial Intelligence) Platform powers financial services digital assistants and chatbots using Kasisto’s industry leading artificial intelligence technology, coupled with extensive financial services domain knowledge and industry conversational data.

The KAI Platform supports multi-modal (text, speech, touch) conversational virtual assistants capable of answering user questions and inquiries, servicing a variety of user requests, and generating leads for new products and services offered. Its natural language understanding capabilities integrate various machine learning, statistical methods and custom rules to most accurately determine the user's intent.

The KAI Platform is composed of multiple components, including;

• Natural Language Understanding (NLU) component that transforms users’ raw input (conversation) into a semantic representation, usually consisting of Intents and Slots, which can then be consumed by the Reasoner, as defined below.

• Reasoner component that determines the next steps that should be taken in the interaction, given understanding of user input. It is the Reasoner’s role to decide to make a call to a backend service, ask the user for more information, or generate a response (or all of the above).

• Natural Language Generation (NLG) component that takes a semantic representation created by the Reasoner and generates text or other (e.g. model) output to be sent to the user in natural language.

• Machine Learning Library (MLL) component that, as KAI Platform processes information, identifies and accumulates new and varied words and phrases, and continuously improves how it associates these to intents and actions. Through ongoing enrichment, quantitative analysis, machine learning, and human inputs, KAI’s MLL has become one of the most robust natural language libraries for human digital experiences in the banking and finance sectors.

The KAI Platform also includes the following APIs and SDKs, which make it simple for developers to interface with KAI and get started quickly:

● The KAI Enterprise API includes definitions and protocols for integrating KAI with a firm’s backend systems, in order to provide users with their firm-specific information, such as their account information (as an example).

● The KAI Conversational API provides a platform-agnostic interface for back-and-forth conversations between the user and KAI. The API supports both plain-text and rich media outputs and can be easily consumed with the use of Swagger code-gen tools.

● The KAI Reporting API includes definitions and protocols for accessing information on the KAI Portal for creation of custom reports and dashboards using a firm’s own business intelligence and reporting tools.

● The KAI Intent API is a framework, which enables firms to define, build, and maintain their Intents, to provide the most up-to-date experience to end users.

● The KAI Machine Learning API includes a set of services, which enables firm’s to train their own natural language understanding (NLU) models. These firms are also able to evaluate their NLU models and receive guidance on how the model accuracy can be improved.

● The KAI Live Chat API provides an interface for integrating KAI to a firm’s live chat service facilitating conversation escalation to live agents.

• A strong foundation of unique, human curated and unbiased data
  • 37M+ real-life user inputs have been collected by KAI Virtual Assistants.
  • The dataset grows by ~1.5M inputs each month
  • KAI uses 5M+ human-enhanced utterances for AI training

• Flexible and Rapid Deployment (Single or Multi-tenant)

  • KAI’s platform deployment capabilities offer customers and partners the hosting choices that meet their technical and business needs. From using KAI within a full cloud environment, to hosting on premises, or a bit of both through hybrid deployments, KAI is flexible.
• Enterprise Ready
  • With KAI’s Conversational API (CAPI) engine, enterprise customers can design their own user interfaces and deeply integrate KAI into preexisting digital experiences and corporate applications. KAI is optimized for performance, scalability, security, and compliance for the regulated industries it serves.
• Powerful Tools to Customize, Track and Improve
  • KAI comes with self-service tools making it easy for clients to add, integrate and test new experiences. With KAI’s enhanced Conversational Management System (CMS), business users can have full control over content responses and customize these responses to the needs of their organization.
  • Intent discovery automatically identifies out of scope experiences and helps to improve conversational coverage by categorizing and organizing new interactions for the business user to review and take action on.

• Multi-lingual