Hebbrix is a persistent memory platform for AI agents and applications that gives models long-term, queryable context by combining a 3-tier memory architecture (short/mid/long-term), a 5-layer hybrid search (vector similarity, BM25, knowledge-graph traversal, time-decay, ONNX reranking) and a knowledge graph for temporal reasoning and contradiction detection with sub-50ms retrieval. It provides OpenAI-compatible APIs, streaming responses, automatic context injection, and automated fact/profile extraction from documents, audio, video and conversations so you can plug it into any LLM stack. Use it to make chatbots, agents, and AI apps that remember, reason, and personalize over time without building custom state-management or memory systems.