{"schemaVersion":"agentdepot.memory_hub.v1","asOf":"2026-10-28","stamp":"[⊙_⊙] ASSAY-VERIFIED","title":"Agent Memory, Vector Stores, Temporal Knowledge Graphs & Context Consolidation Hub","positioning":"How to equip an autonomous agent with short-term working state, episodic session history, consolidated semantic memory, and ACID vector retrieval without context rot or $400/mo cluster lock-in.","memoryArchitectureTiers":[{"tier":"01 // WORKING & CHECKPOINT MEMORY","horizon":"Single run / multi-step graph execution (seconds to hours)","recommendedStack":"Upstash Redis / LangGraph Checkpointer / Postgres","primaryJtbdId":"serverless-redis","rule":"Persist tool outputs and step checkpoints outside the prompt so retries resume from the failed node."},{"tier":"02 // EPISODIC & SEMANTIC AGENT MEMORY","horizon":"Cross-session user preferences, facts & past episode lessons (days to years)","recommendedStack":"Mem0 (extraction + conflict resolution) · Zep / Graphiti (bi-temporal knowledge graph) · Letta (self-editing memory blocks)","primaryJtbdId":"agent-episodic-semantic-memory-learning","rule":"Never append raw chat logs forever. Extract atomic facts with timestamps and resolve contradictions when facts change."},{"tier":"03 // HYBRID VECTOR + SQL DOCUMENT RETRIEVAL","horizon":"Domain corpus, codebases, and multi-tenant RAG (10k to 100M+ chunks)","recommendedStack":"Supabase pgvector (default ACID pick) · Qdrant / Turbopuffer / Pinecone (at 50M+ scale)","primaryJtbdId":"postgres-vector","rule":"Start in Postgres (`pgvector` + HNSW + BM25/tsvector) until you prove you need a separate vector cluster."}],"memoryJtbds":[{"id":"agent-episodic-semantic-memory-learning","name":"Agent Episodic & Semantic Memory Consolidation (Cross-Session Learning)","shortName":"Episodic & semantic memory learning","problemStatement":"Extract durable user preferences, domain facts, and past episode outcomes from raw conversation and tool traces so an agent learns across sessions without bloating the context window or poisoning future runs with stale hallucinations.","depotPickProviderId":"mem0","depotPickName":"Mem0","overallScore":7.59,"scoreAsOf":"2026-10-28","evidenceBasis":"editorial prior (deterministic_source_reverified, ±0.5)","runnersUpIds":["zep","letta","supabase-pgvector"],"whenNotTopPick":"Need temporal bi-temporal knowledge-graph edges with explicit validity windows (`valid_from` / `invalid_at`) for changing facts — Zep (Graphiti); need explicit OS-style self-editing core/archival memory blocks managed by the agent via tool calls — Letta; need DIY SQL + vector episodic tables inside existing Postgres — Supabase pgvector.","decideUrl":"https://getagentdepot.com/api/decide?jtbd=agent-episodic-semantic-memory-learning","humanHref":"https://getagentdepot.com/jtbds/agent-episodic-semantic-memory-learning"},{"id":"postgres-vector","name":"Postgres-native vectors (pgvector)","shortName":"Postgres-native vectors (pgvector)","problemStatement":"Keep vectors next to relational data with minimal new systems.","depotPickProviderId":"supabase-pgvector","depotPickName":"Supabase (pgvector)","overallScore":8.64,"scoreAsOf":"2026-10-28","evidenceBasis":"editorial prior (deterministic_source_reverified, ±0.5)","runnersUpIds":["neon","aws-ec2"],"whenNotTopPick":"Corpus/QPS exceeds comfortable Postgres vector range — Qdrant/Milvus/Turbopuffer.","decideUrl":"https://getagentdepot.com/api/decide?jtbd=postgres-vector","humanHref":"https://getagentdepot.com/jtbds/postgres-vector"},{"id":"embeddings","name":"Embeddings for RAG","shortName":"Embeddings for RAG","problemStatement":"High-quality embeddings (+ optional rerank) for retrieval pipelines.","depotPickProviderId":"cohere","depotPickName":"Cohere","overallScore":7.95,"scoreAsOf":"2026-10-28","evidenceBasis":"editorial prior (deterministic_source_reverified, ±0.5)","runnersUpIds":["voyage","openai","google-ai","jina","huggingface-inference"],"whenNotTopPick":"Already standardized on OpenAI embeddings operationally; retrieval-quality bake-off winner is Voyage; need local/open embed models on HF/Fireworks; want URL-to-markdown + embeddings together — Jina.","decideUrl":"https://getagentdepot.com/api/decide?jtbd=embeddings","humanHref":"https://getagentdepot.com/jtbds/embeddings"},{"id":"postgres-vector","name":"Postgres-native vectors (pgvector)","shortName":"Postgres-native vectors (pgvector)","problemStatement":"Keep vectors next to relational data with minimal new systems.","depotPickProviderId":"supabase-pgvector","depotPickName":"Supabase (pgvector)","overallScore":8.64,"scoreAsOf":"2026-10-28","evidenceBasis":"editorial prior (deterministic_source_reverified, ±0.5)","runnersUpIds":["neon","aws-ec2"],"whenNotTopPick":"Corpus/QPS exceeds comfortable Postgres vector range — Qdrant/Milvus/Turbopuffer.","decideUrl":"https://getagentdepot.com/api/decide?jtbd=postgres-vector","humanHref":"https://getagentdepot.com/jtbds/postgres-vector"}],"benchmarksAndStandards":[{"name":"LongMemEval","url":"https://github.com/xiaowu0162/LongMemEval","focus":"Tests information extraction, multi-session reasoning, temporal updates, and abstention over long chat histories."},{"name":"LOCOMO (Long-Term Conversational Memory)","url":"https://github.com/snap-research/locomo","focus":"Multi-session episodic recall and temporal event ordering across hundreds of turns."},{"name":"MTEB (Massive Text Embedding Benchmark)","url":"https://huggingface.co/spaces/mteb/leaderboard","focus":"Retrieval, reranking, and clustering accuracy across embedding models (`voyage-3`, `text-embedding-3-large`, `bge-m3`)."}],"links":{"self":"https://getagentdepot.com/api/memory","humanPage":"https://getagentdepot.com/memory","swapRecipe":"https://getagentdepot.com/api/swap?from=pinecone&to=supabase-pgvector","learningHub":"https://getagentdepot.com/learning"}}