Native Integration with Amazon Titan Embeddings and More with Cognigy.AI v4.93

With this release, you can now leverage Amazon Titan Embeddings LLM for Knowledge AI in just a few simple steps.

Expanded Standard Embedding Models for Knowledge AI

In line with our goal of delivering the most extensive access to best-of-market LLMs, Cognigy.AI now comes with native integration for Amazon Titan Text Embeddings v2. Working in over 100 languages, this model can accommodate up to 8,192 input tokens and output a vector of 1,024 dimensions.

The latest addition complements existing support for OpenAI's and Azure OpenAI's embedding models, giving Knowledge AI users more flexibility when evaluating and selecting their preferred LLM for knowledge ingestion and processing.

See the complete overview of Cognigy's LLM support by feature here.

Other Improvements

Cognigy.AI

Extended the list of English utterances allowed for the Resolve Phonetic Counters option in the Answer Preprocessing section of the Question Node
Enhanced recognition for voice and chat AI Agents by enabling the use of the article the, as in Three times the four or twice the zero

Webchat

Added the Watermark URL parameter to the Watermark section in the Webchat v3 Endpoint
Moved the Webchat Plugins settings to the Demo Webchat Settings section in the Webchat v3 Endpoint
Updated Webchat v3 to v3.10.0

Cognigy Insights

Added the Goals Performance (in numbers) chart to the Goals dashboard
Shortened the default date range in the Insights global filters from 30 days to 7 days

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