Show HN: Shivvr – Ephemeral semantic embedding and cognitive agent service (shivvr.nuts.services)

Shivvr has launched an innovative ephemeral semantic embedding and cognitive agent service, leveraging advanced AI techniques to enhance text processing and retrieval. This new platform small chunks text and uses GTR-T5-base embeddings for efficient semantic search, allowing users to conduct searches through a hybrid model that combines vector rankings with traditional BM25 scores. The service operates via a distinctive turn-based cognitive agent loop, supporting integrated memory search, document ingestion, and real-time command execution, all while utilizing the Model Context Protocol (MCP) for easy integration with existing AI workflows. This development is significant for the AI/ML community as it highlights the growing trend towards ephemeral computing and enhanced contextual understanding in AI-driven applications. The use of techniques like Reciprocal Rank Fusion (RRF) for blending dense and sparse data, along with secure temporal data management, showcases a new approach to optimizing search relevance and operational efficiency in cognitive systems. With a focus on preserving data privacy through in-memory encryption and a robust API infrastructure for session management and interaction, Shivvr stands to influence the future of intelligent agents and semantic search in various AI applications.