Scalendo's latest Conversation AI release introduces enhanced knowledge base capabilities with three key improvements:

Major Features:

Expanded Data Support: The system now accepts tables (CSV/TSV), rich-text documents, and file uploads (PDF, DOCX, PPT, TXT) alongside plain text, eliminating previous format limitations.

Intelligent Re-Ranking: A new ranking layer evaluates retrieved content snippets for semantic relevance to user queries. This mechanism scores each chunk for semantic closeness to the user's question before generating responses, reducing hallucinations.

Source Attribution: Users can inspect Response Info in conversations to view the exact knowledge chunks, source names, and timestamps powering each AI reply, with quick-edit capabilities for corrections.

Setup Overview:

The implementation involves five steps: creating a knowledge base, adding diverse sources, attaching it to a Conversation AI bot, testing with sample prompts, and reviewing which knowledge chunks informed responses.

Notable Specifications:

• CSV imports support up to 500 columns with selection of the 20 most relevant
• No retraining required after adding new sources
• All data types are indexed instantly and treated equally in retrieval

This update prioritizes response accuracy, faster deployment, and complete transparency regarding AI source material.