Jan 16
Loading video player...
Turn a stateless tool-calling agent into a recall machine. In this short demo, we plug Cognee into LangGraph so your agent can store facts as a knowledge graph and retrieve them across sessions and restarts, grounding answers in real memory instead of guesswork. What you’ll learn - Add a long-term, queryable memory layer to a LangGraph agent - Store structured facts → Cognee auto-builds a knowledge graph - Retrieve and ground answers from prior runs (even after restarts) - Visualize entities & relationships to see how knowledge connects Chapters 0:05 Welcome & goal 0:08 Why Cognee + LangGraph for persistent memory 0:20 What “persistent memory” means for agents 0:41 Install packages 0:53 Keys & imports 1:00 Build the tool-calling agent 1:04 Tools: add (store) & search (retrieve) 1:13 Feed facts → auto-store in Cognee 1:21 Long-term memory workflow 1:32 Knowledge graph (entities & relationships) 1:40 Retrieve & ground answers 2:01 Fresh session, same memory 2:18 Graph visualization 2:51 Community & next steps Try it / Resources Website: https://www.cognee.ai Github: https://github.com/topoteretes/cognee Docs: https://docs.cognee.ai Join the Discord Community: https://discord.gg/cqF6RhDYWz Join Reddit: https://www.reddit.com/r/AIMemory/ LangGraph × Cognee guide: https://www.cognee.ai/blog/integrations/langgraph-cognee-integration-build-langgraph-agents-with-persistent-cognee-memory Integration repo: https://github.com/topoteretes/cognee-integration-langgraph If you ship something with this, tag us - we love seeing agents that actually remember. #langgraph #cognee #aiagents #aimemory #knowledgegraph #llm #langchain #retrievalaugmentedgeneration #GraphRAG #RAG #aiengineering