Cognis: Context-Aware Memory for Conversational AI Agents

TL;DR

Cognis introduces a memory architecture for conversational AI that combines keyword and vector search, enabling persistent, context-aware memory and improved personalization in dialogue systems.

Contribution

A novel multi-stage retrieval pipeline integrating keyword and vector search with temporal and reranking enhancements for persistent conversational memory.

Findings

Achieves state-of-the-art results on LoCoMo and LongMemEval benchmarks.

Demonstrates effective memory retrieval across multiple answer generation models.

System is open-source and deployed in real-world applications.

Abstract

LLM agents lack persistent memory, causing conversations to reset each session and preventing personalization over time. We present Lyzr Cognis, a unified memory architecture for conversational AI agents that addresses this limitation through a multi-stage retrieval pipeline. Cognis combines a dual-store backend pairing OpenSearch BM25 keyword matching with Matryoshka vector similarity search, fused via Reciprocal Rank Fusion. Its context-aware ingestion pipeline retrieves existing memories before extraction, enabling intelligent version tracking that preserves full memory history while keeping the store consistent. Temporal boosting enhances time-sensitive queries, and a BGE-2 cross-encoder reranker refines final result quality. We evaluate Cognis on two independent benchmarks -- LoCoMo and LongMemEval -- across eight answer generation models, demonstrating state-of-the-art performance on both. The system is open-source and deployed in production serving conversational AI applications.