Abstract
Multi-step retrosynthesis planning seeks to decompose a target molecule into commercially available building blocks through a sequence of feasible reactions. The vast combinatorial search space makes this task challenging even for expert chemists. Traditional methods combine tree search with offline-trained value networks that score candidates in isolation, without reasoning about complete multi-step routes. Recent work leverages Large Language Models (LLMs) for this task, but relies on simple interfaces that limit exploration of the full search space. We introduce RetroAgent, an LLM agent that bridges symbolic search and neural reasoning through a harness with structured memory. Through memory and chemistry tools, the agent observes the full search state, including explored routes, available alternatives, and properties of intermediates, enabling informed decisions grounded in both global progress and domain knowledge. Experiments on in-distribution and out-of-distribution benchmarks demonstrate that RetroAgent delivers strong performance and generalization.
Blogger's Review: RetroAgent's design cleverly combines the powerful language processing capabilities of LLMs with the advantages of structured memory, making the complex task of retrosynthesis planning more efficient. With enhanced search state observation capabilities, RetroAgent shows performance that surpasses traditional methods in the field of chemistry, offering broad application prospects. Its strong performance in experiments also provides an important direction for future research.