Bio is a local-first AI research assistant that enables users to explore and analyze distributed biomedical research data in the form of natural language conversations. This tool helps to intuitively integrate and query reliable life science data, such as PubMed scholarly articles, the ClinicalTrials.gov clinical trial database, and the FDA's drug detailed label information, within a single chat interface, streamlining access to information in the medical research field. Similar to how an experienced senior researcher navigates complex clinical data and pharmaceutical information to summarize key reports, Bio organically connects a vast biomedical knowledge database to infer the answers researchers need in real-time. Traditional bioinformatics analysis or literature reviews heavily relied on inefficient manual processes, requiring users to alternate between dozens of data portals, manually copy and paste data, and process it. Furthermore, transmitting sensitive data, such as patient information or undisclosed new drug candidates, to commercial cloud-based Large Language Models (LLMs) poses a significant risk of data leakage. Bio fully supports the integration of local LLMs with offline runtime environments like Ollama and LM Studio, providing a unique security barrier that ensures no data is leaked outside the laboratory's internal workstations or server network. Researchers can use Bio to extract interactions and mechanisms of new anticancer drug candidates from PubMed and quickly summarize the patient recruitment requirements of related Phase 3 clinical trials. In particular, it supports the automatic generation and execution of Python analysis code within the secure sandbox, a segregated code execution environment provided by Daytona, allowing users to effortlessly generate Kaplan-Meier survival curves based on collected patient survival data or perform complex pharmacokinetic mathematical statistical calculations using only natural language requests.