About HumemAI
HumemAI started from Taewoon Kim's PhD, "A Machine With Human‑Like Memory". The core idea: separate conversations from data using human‑inspired episodic and semantic memory, then glue together machine learning, temporal knowledge graphs, and databases to create agentic AI systems with persistent, explainable memory.
Episodic Memory: From conversations to temporal knowledge graphs. Your dialogue with HumemAI agents captures the "where", "when", and "what" of interactions in a time‑aware graph structure. This personal memory evolves continuously, learning from each exchange while maintaining full context of past conversations.
Semantic Memory: Your operational data remains in your databases while we add a thin compatibility layer when needed. Whether structured or unstructured, HumemAI connects via adapters and performs intelligent read/write across your systems. When formats differ, we apply minimal, reversible transformations—not wholesale migrations—to fit supported engines and schemas.
Today, users interact through an LLM while HumemAI performs hybrid retrieval—vector similarity plus relationship‑aware queries over both episodic conversations and semantic data—to ground answers in a continuously evolving knowledge layer. It powers agentic workflows: natural to use, explainable by design, and reusable across tasks.
We're building on open foundations. HumemAI will provide an open‑source SDK that connects to your databases—whether structured or unstructured—while our core engine handles the intelligent memory architecture. Your data stays in your systems; we add the conversation layer as temporal knowledge graphs on top.
Our research focus keeps us ahead: integrating ML with graphs and databases to scale reasoning, improve retrieval, and automate schema/ontology learning. That's how we move beyond short‑lived hype to durable capabilities.
Open Research & Development
HumemAI advances AI memory infrastructure through open‑source contributions and published research. We contribute to open-source databases and share our findings with the community, while building deep expertise in integrating these technologies into production systems.
Our competitive advantage comes from years of experience with temporal graphs, multi‑model databases, and AI architectures — not from hiding code. We believe in building on open foundations while providing the integrated platform and expertise that takes time to master.