Zep is an end-to-end context engineering platform designed to solve the agent context problem. It automatically extracts relationships and maintains a temporal knowledge graph that understands how context evolves over time. The system retrieves pre-formatted, relationship-aware context blocks optimized for LLMs, enabling AI agents to access current, accurate, and relevant information for better decision-making in production environments.