Clinical trials are time consuming, expensive, and often burdensome on patients. Poor protocol design can reduce a success rate of clinical trials. In order to improve protocol design, pharmaceutical companies often spend a lot of time searching for information thoroughly regarding protocol design. In the past years, we have dedicated tremendous efforts to explore and implement novel Artificial Intelligence (AI) applications to advance clinical development. During these projects with Mitsubishi Tanabe Pharma,　we have gained experiences with developing AI applications. Especially, we're focusing on AI applications for protocol design. In these applications, we're leveraging NLP & Deep Learning to help pharmaceutical companies optimize their protocol designs. Our system learns from all kinds of sources, like ClinicalTrials.gov, past clinical trials, journal articles and provide recommendations for clinical trial optimization. This presentation will discuss our early findings on the AI in protocol design.