The recent survey titled "Scaffolded Language Models with Language Supervision for Mixed-Autonomy" provides a comprehensive overview of scaffolded language models (LMs) designed to operate within multi-step processes involving diverse tools. It emphasizes the semi-parametric nature of these models, which blend parametric and non-parametric components to improve performance. A key focus of the survey is on training non-parametric variables, such as prompts and code, to enhance the models' ability to interpret instructions effectively. This approach aims to optimize the integration of language models into complex workflows, supporting mixed-autonomy scenarios where human and machine collaboration is essential. The survey consolidates current research trends and methodologies in this domain, highlighting advancements in model design and optimization. Notably, the findings align with other recent analyses in the field, reflecting a consistent emphasis on semi-parametric architectures and language supervision techniques. Published on arXiv in November 2025, this work contributes valuable insights into the evolving landscape of language model development.