This blog post explores the challenges faced while implementing research workflows using small local large language models (LLMs). It discusses the limitations of processing power, context windows, and the lack of transparency encountered during execution, citing personal experiences and troubleshooting efforts.