Podcast Summary: How I AI
Episode: How this PM streamlines 60k-page FDA submissions and saves millions with Claude, Streamlit, and clever AI workflows | Prerna Kaul
Host: Claire Vo
Release Date: July 14, 2025

Introduction

In this episode of How I AI, host Claire Vo welcomes Prerna Kaul, a seasoned Product Manager with extensive experience in machine learning and AI-driven solutions at prominent companies like Amazon, Moderna, and Panasonic. Prerna shares her innovative approach to utilizing AI—specifically Anthropic's Claude and Streamlit—to streamline the creation of extensive FDA submissions, significantly reducing time and costs while accelerating the delivery of life-saving vaccines and treatments.

The Challenge of FDA Submissions in Life Sciences

Prerna begins by outlining the monumental task her team faced while working on a cancer vaccine at Moderna. The process required the development of a 60,000-page Biological License Application (BLA)—a document critical for FDA approval.

Prerna Kaul [00:15]: “We had to develop a nearly 60,000 page document would have taken about four to six months of effort and nearly 20 specialists. Not to mention the millions of dollars spent.”

The complexity and regulatory requirements in the life sciences industry necessitate meticulous documentation, which traditionally demands significant time and resources. Prerna emphasizes that delays in this process can postpone the availability of crucial vaccines and treatments, directly impacting public health.

Leveraging Claude and AI Workflows for Document Generation

To address this challenge, Prerna leveraged Claude, a generative AI model developed by Anthropic, to automate and optimize the document creation process. She describes her initial approach:

Prerna Kaul [00:27]: “I gave Claude the problem statement, the pitch and a demo. ... It created a little setup, instructions, markdown file for me and it gives me all of the setup instructions and all of the tasks that it does and it keeps capabilities which is super handy.”

By treating Claude as a collaborative software engineer, Prerna was able to generate the foundational code and instructions necessary to build a production-ready system for generating the BLA. Claude not only provided the technical setup but also identified necessary components such as XML formatting and Phi (Protected Health Information) redaction, ensuring compliance with regulatory standards.

Technical Implementation: From Code to Streamlit App

Prerna further details the technical steps taken to transform Claude's output into a usable tool for her team. After reviewing and validating the generated Python script, she deployed it using Streamlit, a platform that facilitates the creation of user-friendly web applications for machine learning and data science projects.

Prerna Kaul [12:18]: “So I kind of started here and started thinking through what do I really need to do to make this a production ready system.”

The Streamlit app enabled non-technical stakeholders to interact with the system seamlessly. Prerna highlights the app’s functionality, which includes generating synthetic clinical data, detecting and redacting Phi information, and compiling these into the required BLA format.

Prerna Kaul [14:04]: “It takes me here. It has preloaded an API key from Claude, which I provided in the code. ... I'm going to generate synthetic data for us.”

Impact: Cost and Time Savings, Life-Saving Implications

The implementation of Claude and Streamlit had profound impacts on both operational efficiency and financial savings. Prerna shares the tangible benefits realized by her team:

Prerna Kaul [24:56]: “We found that cost savings was definitely one thing, but the stakeholders themselves were super engaged in the process and they built a system that scales pretty well as the company scales as well.”

By automating the creation of the BLA, Prerna's team saved months of work, millions of dollars, and reduced the need for a large team of specialists. More importantly, these efficiencies translated into faster deployment of vaccines, potentially saving countless lives.

Additional Workflow: Managing Stakeholders with AI

Expanding beyond document generation, Prerna introduces another AI-driven workflow aimed at product management and stakeholder communication. Utilizing Claude along with literary insights from authors like Jane Austen and Dale Carnegie, she developed a Prompt Generator to assist in navigating complex stakeholder dynamics.

Prerna Kaul [28:47]: “It's a very structured prompt. ... and how it would break down the problem, how it would think and reason through it, and then what outputs it would give me in return as well.”

This tool helps Prerna prepare for meetings by generating structured strategies, communication plans, and even anticipating challenging questions. It serves as a virtual brainstorming partner, enhancing her ability to manage diverse stakeholder expectations effectively.

Addressing Safety, Privacy, and Ethics in AI Workflows

Given the sensitive nature of health data, Prerna places significant emphasis on safety, privacy, and ethical considerations in her AI workflows. She outlines a structured approach to ensure compliance and ethical alignment:

Prerna Kaul [41:16]: “Compliance and privacy comes first. But it's a bit more than privacy and compliance. It's the ethics overall that one must prioritize.”

Prerna collaborates closely with her team to select AI models that prioritize safety and alignment with human values. She advocates for continuous monitoring and evaluation of AI performance both offline and in production to maintain high ethical standards.

Conclusion and Final Thoughts

Prerna Kaul's innovative use of Claude and Streamlit exemplifies how AI can revolutionize complex, resource-intensive processes in the life sciences industry. By automating FDA submissions, her approach not only saves substantial time and money but also accelerates the delivery of life-saving medical treatments. Additionally, her AI-driven strategies for stakeholder management highlight the versatile applications of generative AI in product management.

In her closing remarks, Prerna encourages her peers in life sciences and other regulated industries to embrace AI tools to optimize their workflows and enhance their impact.

Prerna Kaul [39:27]: “I would encourage people in life sciences ... to play around with these tools and not be afraid of actually trying to optimize their own work because they might find it's a common thing for all of their peers in the process.”

Notable Quotes

• Prerna Kaul [00:15]: “We had to develop a nearly 60,000 page document would have taken about four to six months of effort and nearly 20 specialists. Not to mention the millions of dollars spent.”
• Clarie Vo [10:07]: “If you're getting internal resistance to cost... bring true transparency and ... sense of ROI and investment ...”
• Prerna Kaul [41:16]: “Compliance and privacy comes first. But it's a bit more than privacy and compliance. It's the ethics overall that one must prioritize.”
• Prerna Kaul [39:27]: “I would encourage people in life sciences ... to play around with these tools and not be afraid of actually trying to optimize their own work.”

Prerna Kaul's episode on How I AI serves as a compelling case study on the transformative power of AI in highly regulated and complex industries. Her methodologies offer valuable insights for professionals seeking to integrate AI into their workflows to achieve significant efficiency gains and impactful outcomes.