AI and the World's Hard Problems · Chemistry & materials · arXiv · 2026-10-21
Magnolia: A Persistent-Memory Agent Framework for Computational Chemistry
As chemistry enters a high-data regime, dedicated scientific agents are becoming central to guiding chemists through ever-increasing data and findings. The study introduces Magnolia, a persistent-memory AI agent framework that integrates computational protocols from docking to quantum chemistry with lab notebooks and chat logs through large language models, using a three-tier memory system in which validated learnings are promoted to cross-project rules. Across four case studies it tracked 317 runs and accumulated 116 project-tier entries, and in controlled replays runs with memory passed checks that runs without it failed, such as ten of ten docking-configuration checks against one to four of ten.