AI ENGINEER SHANGHAI 2026
Rentong Guo
Co-founder & VP of Product
Zilliz
Talk
Your Agent's Memory Is Next Quarter's Training Data
November 5 · 15:55-16:15 · Main stage →
Shanghai time (UTC+8) · Changes will be communicated.
An agent that runs twenty steps doesn't issue one retrieval — it issues dozens, and it writes about as much as it reads. The trace it leaves behind is write-heavy, time-ordered, and queried at three granularities: step, span, trajectory.
That's the same shape as an RL rollout store. Not similar to — the same. The model labs hit this two years earlier under different names: semantic dedup, decontamination, hard-negative mining.
Agent memory and training data are converging on one set of operators — operational writes and analytical scans over the same data, with no ETL in between. That convergence is why we took Milvus from a vector database to a Vector Lakebase. This talk walks through the workloads that forced the shift, and what the architecture looks like underneath.
About the speaker
Rentong Guo is Co-founder and VP of Product at Zilliz and a core architect of Milvus. He holds a PhD from Huazhong University of Science and Technology and has extensive experience in large-scale AI data infrastructure and distributed systems. He is advancing data infrastructure for the next generation of agents and AI applications.