Researchers propose a framework for AI systems to serve as personalized co-scientists by incorporating individual researcher representations into every stage of the research process. This approach aims to address the limitation of current state-of-the-art systems that prioritize novelty, validity, or reviewer score over the specific needs and context of each researcher. The framework consists of three components: graph-grounded researcher representations, personalization across the full research pipeline, and evaluation grounded in the individual. --- Why it matters: This matters because AI co-scientists could revolutionize research by automating tasks such as hypothesis generation, experiment design, and paper writing. However, their effectiveness depends on their ability to adapt to individual researchers' needs and context, which is currently lacking. This work addresses this limitation and has the potential to make AI co-scientists more useful and effective tools for researchers. Source: https://arxiv.org/abs/2608.14881