Memory · Data store

Episodic Memory Store

Data storeMemoryCognition & Memoryarc:EpisodicMemoryStore

Long-term memory of specific timestamped past events and interactions, queried by temporal proximity and event attributes.

Responsibility. Stores past interactions and events chronologically.

Also known as: Episodic memory, Interaction history, Long-term memory, Long-term graph memory, Long-term episodic memory, Experience timeline, Episode store, Trajectory memory, Shared fleet episodic memory, Conversation archive, Previous-session summaries

Direct neighbourhood (hover for relationship types)

Relationships

is guarded by control

Memory Write Validator Ch1.4 Ch3.5

Design guidance

MUST NOT be implemented solely on a vector store; vector similarity cannot preserve chronology or detect repeats.
SHOULD be backed by relational or time-series storage that supports temporal queries and aggregation.
SHOULD persist durably across agent restarts and support similarity-based retrieval via vector embeddings.
SHOULD store metadata such as confidence scores, relevance indicators and temporal information alongside memories.
SHOULD keep historical interactions in long-term memory retrieved on demand rather than in the prompt.
SHOULD use a precise semantic indexing strategy so retrieval does not return excessive context that inflates tokens.
SHOULD persist episodes in an external store rather than the context window so history scales beyond session and window limits.
SHOULD record episodes as structured representations (state, actions, outcomes, context metadata, agent state, importance signals) rather than raw transcripts.
SHOULD attach freshness metadata (timestamps, duration estimates, confirmation status) and update shared episodes as conditions change.
SHOULD isolate per-user episodic memories (e.g., customer-specific namespaces) when comparisons must reflect personal rather than global patterns.
SHOULD NOT be treated as a generalization mechanism; storing episodes is memorization, and generalization requires explicit abstraction into semantic or procedural memory.

Quantitative guidance

As stated by the sources; verify before use.

Episodic memory in external databases scales to millions of past interactions vs. 100K-200K-token context windows (Ch5.7).
Episode embeddings typically 768-1536 dimensions (Ch5.7).
Fraud example: 90-day per-customer profile build-up before anomaly event on day 120 (Ch5.7).

Classification

Patterns: Temporal filteringFew-shot learning from historyShort-term/long-term memory splitTiered memoryStructured episode anatomy (state, actions, outcomes, context metadata, agent state, importance signals)Trajectory storage as (state, action, observation) sequences with trajectory-level metadataTiered storage (detailed high-importance episodes, aggressively summarized routine episodes)Per-customer namespacesShared multi-agent episodic memory with in-place episode updatesVector storage with metadata filtering; optional graph layer for episode relationshipsStore for intermediate per-pass extractions and comparison matrices in multi-pass processingEvent logs with timestamps (Ref5.04)
Technologies: SQLiteRelational databaseTime-series databaseNeo4jPineconeWeaviateChromaAmazon Neptune
Quality attributes: Security (ISO/IEC 25010 | NIST AI RMF: secure and resilient)Interaction capability (ISO/IEC 25010)Reliability (ISO/IEC 25010 | NIST AI RMF: valid and reliable)Functional suitability: correctness and validity (ISO/IEC 25010 | NIST AI RMF: valid)Performance efficiency (ISO/IEC 25010)Flexibility (ISO/IEC 25010)
Risks mitigated: Action duplicationUnresolved implicit referencesLoss of cross-session context (users repeating themselves)Stateless agents unable to learn from experienceContext-window capacity limits on history

Sources

Ch1.4: T. Nguyen, "Memory and Perception Systems," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 1.4. ISBN: 9798244538229.
Ch1.5A: T. Nguyen, "Stateful Orchestration - Introduction and Core Concepts," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 1.5A. ISBN: 9798244538229.
Ch1.6: T. Nguyen, "Stateful Orchestration - Pitfalls, Integration, and Synthesis," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 1.6. ISBN: 9798244538229.
Ch1.7B: T. Nguyen, "Relational Reasoning with Knowledge Graphs - Hybrid RAG+KG Integration," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 1.7B. ISBN: 9798244538229.
Ch3.10: T. Nguyen, "Efficiency Metrics," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 3.10. ISBN: 9798244538229.
Ch5.1: T. Nguyen, "Chain-of-Thought (CoT) Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.1. ISBN: 9798244538229.
Ch5.2: T. Nguyen, "Tree-of-Thought (ToT) Fundamentals," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.2. ISBN: 9798244538229.
Ch5.7: T. Nguyen, "Episodic Memory," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.7. ISBN: 9798244538229.
Ch5.8: T. Nguyen, "Semantic Memory," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.8. ISBN: 9798244538229.
Ch5.9: T. Nguyen, "Working Memory," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 5.9. ISBN: 9798244538229.
Ch10.1: T. Nguyen, "Conversational UI," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.1. ISBN: 9798244538229.
Ch10.2: T. Nguyen, "Proactive Agents," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 10.2. ISBN: 9798244538229.
Ref5.04: C. Stryker, "What is AI agent memory?," IBM Think. Accessed: Sep. 27, 2026. [Online]. Available: https://www.ibm.com/think/topics/ai-agent-memory