Intelligence That Makes Chatbots More Helpful and Human
Develop chatbots that can contextualize, remember, and learn from conversation data to execute actions and make relevant future suggestions.
Chatbots that understand and use contextual information throughout conversations will interact in a way that’s easier, quicker, and more naturally helpful for people. However, information collected from users, your company, and an individual bot will vary in importance, utility, and lifespan. Some data is time sensitive. Some needs to be carried across conversations. Some is static, and some is dynamic.
Make smarter chatbots that can leverage relevant data in the right circumstances
Kore.ai’s Platform supports multiple context variables to allow for thoughtful design of how this information gets used for distinct scenarios. Through the Bot Builder tool, developers can customize, categorize and apply contextual information that maximizes the chatbot’s ability to make intelligent decisions and avoid user confusion or extra steps.
Bots remember actions, data, and contextual details to maintain conversation continuity
Enterprise systems can maintain and remember customer, employee and company specific information. But that information is static, contained within a single location, and rarely remembered in context for exactly the right amount of time. With Kore.ai’s Platform, developers can also define when the chatbot will remember and use contextual data – either in the short term or long term – to best meet enterprise objectives, steer intelligent dialogs, and promote user convenience.
Kore.ai’s NL engine can also detect the sentiment of an interaction to help steer the flow of the conversation
Beyond completing tasks, Kore.ai Bots can also understand a user’s mood throughout a conversation. Our NLP engine scores sentiment based on connotation, word placement, and modifiers. Developers can use these scores to trigger custom flows to improve bot-to-user communication or bring in human agents as needed.
Kore.ai analyzes six possible emotions – anger, disgust, fear, sadness, joy, and positivity. The sentiment algorithm can also score for multiple emotions. For example, an input could yield a high score for joy, but a mild score for sadness.
Our NL engine scores sentiment on a scale of -3 to +3 – with positive values representing expressed emotions, and negative values representing suppressed emotions.
The overall tone score is calculated and assessed for modifiers for example, “I am extremely disappointed” would return a higher angry tone than, “I am disappointed.” Developers can store these tone scores as context and alter the flow of conversations using the Dialog Builder within Kore.ai’s Bot Builder.
The Kore.ai Bots Platform facilitates learning as a component of our natural language engine. Since every interaction gets logged and categorized as a success or failure, bots can be designed to learn from those lessons – both positive and negative – and be adjusted accordingly through human intervention or automatically.
Continuously review, improve, and perfect your bot’s natural language
Supervised learning for NL aids bot intelligence by allowing a human to analyze and respond to the way people communicate with the bot once it’s up and running. Through our Bot Builder tool, developers and admins can evaluate all interaction logs, easily change NL settings for failed scenarios, and use the learnings to retrain the bot for better conversations. Developers can also leverage chat logs to build predictive models and use the outcomes to further define additional proactive alerts, suggested actions, or automated workflows.
How it is implemented
Automatically multiply your bot’s language model when appropriate
Unsupervised learning for NL can be applied to expand the language capabilities of your chatbot – without human intervention. Unlike unsupervised models in which chatbots learn from any input – good or bad – the Kore.ai Bots Platform enables chatbots to automatically increase their vocabulary only when the chatbot successfully recognizes the intent and extracts the entities of a human’s request to complete a task.