.png)
Moderna
Life Sciences · US
50% faster clinical trials
Moderna deploys 750+ custom GPTs across 3,000 employees and halves clinical trial protocol generation time
Rather than piloting GenAI in one team, Moderna implemented ChatGPT Enterprise at company-wide scale and built 750+ internal custom GPTs in eight months, including Dose ID for data-driven dose selection in clinical trials.
Friction
Moderna's pivot from COVID vaccine economics to a broader oncology and rare disease pipeline required dramatically faster time-to-market with strong cost discipline. Clinical trial protocols alone took 6+ months per programme to generate manually.
Breakthrough
ChatGPT Enterprise with data isolation and 750+ custom GPTs built on internal SOPs, clinical trial history, and regulatory templates. Custom GPTs have specific instructions, retrieval over approved documents, and in some cases tools such as dose calculation. Daily audit trails for regulatory compliance.
Impact
Clinical trial protocols now generated in 50% of the previous time via a custom GPT processing SOPs and historical trial designs. 80% of employees use ChatGPT Enterprise weekly. Dose ID supports the clinical team in data-driven dose selection, previously days of manual work. CEO Stéphane Bancel publicly committed AI as core to Moderna's strategy to launch 15 products in 5 years.
Unlock the full analysis with breakthrough, impact, what made it smart and its technical approach below!
Problem
Bringing a new drug to market requires an enormous volume of structured, precise documentation: trial protocols, regulatory filings, safety reports. Each document takes months when written from scratch. Moderna needed to multiply its pipeline without multiplying its workforce proportionally.
What made it smart
The custom GPT architecture allowed Moderna to encode domain expertise into the AI system itself. Rather than a general-purpose assistant, each custom GPT is a specialist trained on the specific documents, SOPs, and decision rules for its task domain, making the output immediately usable rather than requiring extensive human revision.
Technical approach
ChatGPT Enterprise deployed company-wide with data isolation per user and workspace. Custom GPTs built on internal SOPs, historical trial designs, and regulatory templates. Retrieval over approved document sets ensures outputs are grounded in Moderna's own validated knowledge base. Audit trails support regulatory compliance.
Strategic lesson
The multiplier effect of GenAI comes from domain specificity. A custom GPT trained on your SOPs and history is not just faster than a general assistant, it produces outputs that are immediately usable in your regulated workflow.
Reflection question
Where in your organisation are expert documents such as protocols, standards, templates, and historical decisions sitting in a database that no one has time to read before starting a new project?