How does the RIFF framework function?

RIFF functions by transforming content into interactive knowledge through a four-step process. It begins with knowledge ingestion, where documents and data are uploaded and synthesized into a unified knowledge graph. Key steps include: • Knowledge Ingestion: Upload docs, PDFs, or connect to Google Drive/Notion. RIFF chunks content, generates embeddings, and stores it in PostgreSQL with pgVector. • Knowledge Synthesis: Validates new content against a Golden Corpus, classifies claims, resolves conflicts, and auto-generates FAQs. • Deployment: Offers conversational interfaces on websites, streaming responses in real-time and supporting multimodal content. • Continuous Improvement: Learns from conversations, captures feedback, suggests training improvements, and maintains accuracy. RIFF's multi-tier retrieval system prioritizes trained responses and leverages a hierarchy of authority for content validation.