About HumemAI
HumemAI was born from Taewoon Kim’s PhD research, where he explored the intersection of knowledge graphs, machine learning, and human cognition to create AI that can remember, learn, and adapt over time.
Unlike traditional AI models that start every interaction from scratch, HumemAI introduces a memory-driven approach. It automatically constructs a domain-specific ontology to narrow down the search space for data, converting unstructured information into a structured knowledge graph stored in a graph database.
By integrating LLMs, users can interact naturally without dealing with complex query languages—they can simply ask questions, and HumemAI retrieves the most relevant insights from its evolving knowledge base.
This system is designed for business intelligence, research automation, and AI assistants that need long-term learning capabilities. From analyzing large datasets to providing context-aware AI solutions, HumemAI is pushing the boundaries of what AI can remember and understand.
Our long-term goal is to develop the most general ontology capable of covering any dataset. However, in many practical applications, such a broad ontology would be overkill. That’s why HumemAI dynamically generates specific ontologies tailored to each problem. This process is semi-automated with minimal human intervention, ensuring efficiency while maintaining accuracy.
The HumemAI team has a strong academic research background, pioneering not only applied AI solutions but also contributing to fundamental research in AI, knowledge graphs, and cognitive computing. We are dedicated to advancing the field, bridging the gap between theoretical research and real-world AI applications.