This paper introduces data-to-paper, an innovative automation platform that uses interacting Large Language Model (LLM) agents to conduct end-to-end scientific research. The system mimics human scientific practices by raising hypotheses, designing research plans, executing code, interpreting results, and producing complete, traceable research papers. Data-to-paper operates in both autopilot and co-pilot modes, demonstrating its ability to generate insights from data autonomously, although human assistance becomes vital as complexity increases. A key feature is the "data-chaining" which links results to their source code, enhancing verifiability. The research underscores the potential of AI in accelerating scientific discovery while maintaining transparency and traceability, while also noting the need for humans to be co-pilots to these technologies.