Expert-guided Bayesian optimization for sustainable protein formulation

Autonomous scientific discovery systems increasingly use LLMs to narrow design spaces before experiments are run, but this practice is double-edged: when the LLM is right, sample efficiency can improve dramatically; when it is wrong, the system can underperform random search. We formalize Expert-Guided Bayesian Optimization (EGBO), in which an expert, e.g. a human or LLM, selects a low-dimensional subspace for BO and may adaptively expand it over time. We decompose EGBO's suboptimality into a selection gap and an optimization gap, and characterize the coverage–dimension tradeoff governing when expert guidance helps. To support in silico prototyping before costly real-world deployment, we introduce FormulateBench, a suite of 24 plant-based formulation tasks, on which LLM-guided EGBO outperforms all tested baselines. When deployed to optimize two plant-based dairy products, EGBO improves utility, as assessed by a trained human panel, by 29\% and 26\% in 10 iterations each. In a comparison with a professional human food scientist given the same time budget, EGBO achieved near-perfect utility of 0.992, vs. 0.850 for the food scientist.