Enterprise conversational systems require memory to personalize interactions, retain information across sessions, and maintain conversational coherence. At enterprise scale, however, memory must not only retrieve relevant information accurately but also provide it efficiently to the answer-generation model. Existing memory systems face a trade-off between retrieval accuracy and context efficiency. Retrieving more context can improve recall but increases inference cost and latency. We present Scalaris, a scalable enterprise memory architecture that addresses this trade-off through a unified long-term and short-term memory design. We propose that user interactions and assistant responses possess distinct characteristics and roles in a conversation: user interactions capture evolving knowledge about the user and provide value across sessions, whereas assistant responses must be stored efficiently so that only information relevant to the current query is retrieved. We therefore treat them separately and support each with specialized representations and ingestion/retrieval pipelines. Scalaris reaches 92.2 % accuracy on LongMemEval-S and 90.77 % on LoCoMo using Qwen3-32B, state-of-the-art results on both benchmarks, while supplying the answer-generation model with approximately 10 times fewer retrieved context tokens than Hindsight, a leading memory system for conversational AI. When integrated with our proprietary closed-source AI assistants, Scalaris improved overall response quality by 36 %. Together, these results establish Scalaris as a cost-effective framework for building scalable, personalized conversational AI systems.
Figures & tables
Figure 1: Overview of Scalaris’s source-aware memory organization. User interactions are processed to maintain relevant facts, relevant entities, and user and session summaries in a temporal knowledge graph. Assistant responses are decomposed into independently retrievable atomic units and stored in a vector store. In parallel, both user interactions and assistant responses are preserved verbatim in a shared history store.
Figure 2: An illustrative instance of context retrieval in Scalaris. Long-term memory supplies the user’s preference for fuel economy together with prior Honda Civic interactions, while short-term memory supplies the Relevant Assistant Messages, session summary, and recent turns (only the latest exchange is shown, for visual clarity). Together these resolve the referent, let the agent determine that only Corolla information must be retrieved, and enable personalization by focusing on fuel economy and including trade-offs.
Figure 3: Flow diagram of the ingestion and retrieval pipeline of Scalaris. The ingestion phase organizes user interactions as facts, entities, and summaries in the graph store, assistant responses as atomic units in the vector store, and stores each incoming user query and AI response verbatim in the history store. The retrieval pipeline begins by reformulating the incoming query, which is then used to retrieve information from the graph and vector stores. Summaries are fetched directly from the graph store, and the N most recent exchanges from the history store. All retrieved components are combined during context assembly and passed to the answering system.
Table 2: Performance comparison on LoCoMo (% Accuracy). S-Hop: Single-Hop; M-Hop: Multi-Hop; O-Domain: Open Domain.
Method
LME Acc.
LME
LoCoMo Acc.
LoCoMo
(%)
Tokens
(%)
Tokens
Hindsight
94.6
43,625
92.0
36,235
Scalaris
92.2
4,116
90.77
4,145
Table 3: Tokens vs. accuracy (LME = LongMemEval).
Memory organization
LongMemEval-S
LoCoMo
User interactions
Assistant responses
Acc.
Avg. tokens
Acc.
Avg. tokens
Graph
Graph
76.2
5,532
73.5
5,729
VectorDB
Graph
79.3
6,267
77.2
6,183
VectorDB
VectorDB
86.3
8,729
88.5
8,325
Graph
VectorDB
92.2
4,116
90.77
4,145
Table 4: Ablation study of memory organization on LongMemEval-S and LoCoMo (% Accuracy). The last row is the proposed Scalaris organization.
Appendix figures & tables4 assets
Supplementary material from the paper’s appendix.
Appendix
Retrieval Method
k=5
k=10
k=15
k=20
k=25
Semantic Search
52.1
62.3
69.2
69.7
69.9
Lexical Search
44.1
49.2
54.7
55.2
55.3
Symbolic Search
36.7
37.7
37.8
37.8
37.8
Appendix
Table 5: Retrieval analysis for Relevant Assistant Messages (Single-Session-Assistant). The optimal retrieval budget for each method (the elbow point beyond which gains become negligible) is shown in bold .
Retrieval Method
k=5
k=10
k=15
k=20
k=25
Semantic Search
63.8
71.9
77.2
79.6
79.8
BM25
59.4
66.8
72.5
72.9
73.0
Semantic + BFS
74.6
80.9
84.2
85.8
85.9
BM25 + BFS
67.5
74.8
79.8
80.3
80.4
Appendix
Table 6: Retrieval analysis for Relevant User Interactions (Single-Session-User). The optimal retrieval budget for each method (the elbow point beyond which gains become negligible) is shown in bold .
Hyper-
Range
Optimal
Eval.
parameter
Evaluated
Range
Dataset
Semantic similarity threshold
0.3–0.7 (step 0.1)
0.4–0.6
LME
BM25 threshold
0.6–1.1 (step 0.1)
0.9–1.0
LME
Graph traversal depth
1–5 hops
3 hops
LME
Last- N Exchanges
1, 3, 5, 7, 10
5
Internal Eval.
Appendix
Table 7: Evaluated ranges and recommended values for Scalaris’s primary hyperparameters. LME: LongMemEval.
Figure 4: Evaluation prompt used by the GPT-4o judge to compare two AI assistant responses. The prompt provides the original user question, recent conversational history, and available memory context, followed by both candidate responses. The judge assesses answer quality, personalization, clarity, actionability, and consistency with the conversation before assigning a score from 0 to 5 to each response and producing a comparative verdict in a strictly specified JSON format.
Conversational memory equips AI assistants to retain, update, and retrieve useful information from past interactions for coherent, personalized dialogue.
Efficient long-term conversational memory requires retrieving sufficient evidence without indiscriminately expanding the context presented to the language model. This is challenging because relevant evidence may be distributed across multiple sessions, while compression may discard details needed for answering. Different queries therefore require different forms of memory access. To capture these demands, we formulate memory access along two dimensions: discovery breadth, which controls how broadly evidence is searched, and reading fidelity, which controls whether evidence is read in compact form or recovered from the original conversation. Based on this formulation, we introduce JustMem, which stores conversation history as compact atomic memories and adapts memory access along these two dimensions to each query. Specifically, LOOKUP handles local evidence, COMPOSE broadens discovery for distributed evidence, and REPLAY increases reading fidelity for fidelity-sensitive evidence. On LoCoMo and LongMemEval-S, JustMem achieves the highest mean accuracy and retrieval recall among the compared memory systems while using substantially fewer generative-model tokens for memory construction and inference.
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.
Large language model assistants are increasingly expected to retain and reason over information accumulated across many sessions. We introduce EngramaBench, a benchmark for long-term conversational memory built around five personas, one hundred multi-session conversations, and one hundred fifty queries spanning factual recall, cross-space integration, temporal reasoning, adversarial abstention, and emergent synthesis. We evaluate Engrama, a graph-structured memory system, against GPT-4o full-context prompting and Mem0, an open-source vector-retrieval memory system. All three use the same answering model (GPT-4o), isolating the effect of memory architecture. GPT-4o full-context achieves the highest composite score (0.6186), while Engrama scores 0.5367 globally but is the only system to score higher than full-context prompting on cross-space reasoning (0.6532 vs. 0.6291, n=30). Mem0 is cheapest but substantially weaker (0.4809). Ablations reveal that the components driving Engrama's cross-space advantage trade off against global composite score, exposing a systems-level tension between structured memory specialization and aggregate optimization.