LLM-Powered Entity Extraction, Expanded Language Assist in AI Copilot, and More with Cognigy.AI v4.72

LLM-Powered Entity Extraction, Expanded Language Assist in AI Copilot, and More with Cognigy.AI v4.72

5 min read

Authors name: Nhu Ho March 26, 2024


With the latest release of Cognigy.AI, level up the language understanding of your AI Agents to new heights, empower agents with real-time voice translation, and enable proactive Knowledge Assist in AI Copilot.

GenAI-Powered Entity Recognition for Unparalleled Understanding

Traditional NLU systems often rely on predefined ontologies or schemas, limiting their ability to recognize entities outside these parameters. On the other hand, human language, especially in speech, often involves multiple digression paths and subtle nuances. Take the following example:

AI: Please tell me your license plate number.

Customer: It’s B C five four, oh sorry, it’s B C five three eight nine F

Instead of correctly retrieving BC5389F, traditional NLU might misinterpret this as BC5 4BC5 389F or fail to comprehend altogether if the format has been specified.

Cognigy.AI now lets you transcend this limitation by harnessing the remarkable language understanding of Generative AI. With v4.72, we’ve introduced a new LLM-powered Entity Extraction feature, both as a standalone Node and integrated within the Question Node.

Merely by providing a simple description of the entity format and a handful of user input examples, you can leverage Generative AI to boost the entity recognition accuracy of your AI Agents.

LLM Entity Extract | HubSpot Video

Empower Digital Contact Center Agents with Multilingual Voices

Cognigy lets you build multilingual AI Agents to not only enhance self-service, but also augment your workforce. The newest update broadens our Language Assist in AI Copilot by introducing real-time Speech-to-Text-to-Speech translation.

This advancement empowers digital contact center agents by giving them a voice in any language. When calling your hotline, customers can interact with a Voice AI Agent in their native tongue. Upon handover, a service agent can continue the conversation on chat in their preferred language, while the AI seamlessly handles live translation over the phone. Experience it firsthand here.

To implement this, activate Recognize Language for automated language detection in the Set Session Config Node. You can then configure the alternative languages of your choice to set the translation respectively.

Note that different speech providers offer different settings and availability for language recognition. Alternatively, you can leverage LLM for language recognition.

Extended Knowledge Assist in AI Copilot

Another highlight from Cognigy.AI v4.72 is the newly added Knowledge Assist option in the Copilot: Next Action Tile Node. Previously, the Knowledge Tile Node enabled agents to actively search for the information they need within the AI Copilot Workspace.

The new feature complements existing capabilities by autonomously suggesting the next best reply through Knowledge AI during live support.

Other Improvements

Cognigy.AI

Cognigy Live Agent

For further information, check out our complete Release Notes here.