The Center for Data Innovation recently spoke with Chetan Mishra, CEO of Rhizome AI, a New Jersey-based company providing a large language model to help life science teams quickly find and understand the regulatory and clinical information they need. Mishra explained how the company’s model retrieves trusted information from global health regulators to generate traceable, evidence-backed insights that help drug developers make faster, more informed decisions.
David Kertai: What does Rhizome AI do?
Chetan Mishra: Rhizome is a large language model (LLM) chatbot for the life sciences industry. We work with biotech and pharmaceutical companies to help scientists, regulatory teams, and clinical researchers quickly find and interpret the regulatory guidance, precedent decisions, and scientific evidence they need to bring new therapies to market.
Drug developers must review thousands of pages of regulatory guidance and past approval decisions about how similar medicines were evaluated before they plan their own studies, design trials, or submitting applications; and even small errors can delay approval by months or years. Because regulators rely heavily on precedent, companies must understand how regulators evaluated similar products to shape their own strategy.
Biotech and pharmaceutical companies use our LLM to ask regulatory and clinical questions and receive cited answers drawn from verified documents. We pair our LLM with proprietary retrieval and validation architecture that gets the system to display the exact paragraph and source document behind every response. This citation requirement makes each answer traceable and significantly reduces hallucinations.
Kertai: What types of data does your LLM use?
Mishra: Our LLM draws on more than 2.5 terabytes of public data from major health authorities, including the U.S. Food and Drug Administration, the European Medicines Agency, the U.K. Medicines and Healthcare products Regulatory Agency, and the Pharmaceuticals and Medical Devices Agency of Japan, along with global clinical trial registries. We structure regulatory guidelines, approval decisions, and clinical trial data covering study designs, endpoints, and results.
Kertai: What are the biggest challenges you’ve faced in ensuring your LLM’s accuracy?
Mishra: Reliability is essential in life sciences because inaccurate or outdated information can lead to noncompliance, flawed study designs, or costly delays. The biggest technical challenge for us is retrieval: regulatory documents vary widely in structure, formatting, language, and quality, and many appear only as scanned PDFs or contain inconsistent metadata.
To ensure accuracy, we pull guidance documents, approval packages, advisory committee materials, and clinical trial records directly from global health‑authority portals and registries that often update without notice. These materials arrive in many formats, so our pipelines clean, normalize, and rebuild them into a consistent structure. Ingesting complete, verified source text upfront minimizes the risk of outdated or incomplete information shaping regulatory decisions.
Kertai: How is your approach different from other companies?
Mishra: Many organizations still rely on internal teams, consultants, or legacy subscription databases to gather regulatory intelligence. Traditional tools mainly serve as document repositories, they store information but do not connect or reason across it. Some companies focus narrowly on regulatory or clinical writing, which optimizes a single workflow but does not address the broader intelligence layer needed for strategic decisions.
Our LLM targets that intelligence layer. It allows teams to analyze global precedent, compare regulatory decisions across regions, and generate evidence-backed insights in a single system. As AI capabilities advance, we expect regulatory and clinical workflows to shift from manual document retrieval to structured, explainable decision support.
Kertai: What’s your vision for Rhizome’s future?
Mishra: We aim to make complex scientific and regulatory information instantly searchable, explainable, and actionable. Today, teams across research and development, regulatory, clinical, and commercial functions spend weeks or months assembling and analyzing regulatory intelligence for major development decisions.
Our vision is to compress that analytical and documentation work into five days. This goal refers to regulatory analysis, precedent review, and strategic documentation, not the physical execution of clinical trials. By shortening these knowledge cycles, we aim to help companies reduce development timelines and bring more therapies to patients faster.
