# Cognis: Context-Aware Memory for Conversational AI Agents

Source: [Cognis: Context-Aware Memory for Conversational AI Agents](<https://www.sota2.com/research/paper/69e9a76255a3c32af89f6ea4>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-24T05:00:04.605Z

Authors: Parshva Daftari, Khush Patel, Shreyas Kapale, Jithin George, Siva Surendira

Published: 2026-03-27T10:15:30.000Z

## 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.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Long-context Question Answering on LoCoMo Single-Hop 2024](<https://www.sota2.com/research/sota/long-context-question-answering-on-locomo-single-hop-2024.md>): Long-context Question Answering; LoCoMo Single-Hop 2024; F1 Score: 48.66.

- [Long-context Question Answering on LoCoMo Multi-Hop 2024](<https://www.sota2.com/research/sota/long-context-question-answering-on-locomo-multi-hop-2024.md>): Long-context Question Answering; LoCoMo Multi-Hop 2024; F1 Score: 31.51.

- [Long-context Question Answering on LoCoMo Open-Domain 2024](<https://www.sota2.com/research/sota/long-context-question-answering-on-locomo-open-domain-2024.md>): Long-context Question Answering; LoCoMo Open-Domain 2024; F1 Score: 54.77.

- [Long-context Question Answering on LoCoMo Temporal 2024](<https://www.sota2.com/research/sota/long-context-question-answering-on-locomo-temporal-2024.md>): Long-context Question Answering; LoCoMo Temporal 2024; F1 Score: 62.68.

- [Long-context Memory Evaluation on LongMemEval](<https://www.sota2.com/research/sota/long-context-memory-evaluation-on-longmemeval.md>): Long-context Memory Evaluation; LongMemEval; Average Score: 92.4.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2604.19771>)

[PDF](<https://arxiv.org/pdf/2604.19771.pdf>)