1. What AI Is Used in CoAuthor
CoAuthor is powered by an open-source large language model deployed and hosted by Certara within a secure, isolated infrastructure environment. Certara is the deployer of this model we did not create or train the underlying model, and we are not the model provider.
The model operates in a fully air-gapped configuration: it has no internet connectivity and no access to external systems. It does not learn from, retain, or incorporate any customer-submitted data. All model inference occurs within Certara's controlled environment.
AI capabilities are available across two feature categories in CoAuthor:
Prompt-based draft generation — individual section prompts or batch generation across an entire document using the Generate All function.
AI Chat — a conversational interface for document-related queries and content assistance.
2. How AI Is Used
The AI features in CoAuthor generate draft content based on prompts and user-supplied context. All AI output is a starting point for human authoring — it is not submission-ready and is not intended to be treated as such.
CoAuthor AI operates as a drafting assistant. It accelerates the authoring process by generating initial content that medical writers then review, edit, validate, and approve in accordance with their organization’s quality and regulatory requirements. The AI does not make regulatory determinations, interpret data, or produce final deliverables.
3. Human Editorial Responsibility
Medical writers and regulatory authors bear full editorial responsibility for all content produced using CoAuthor. This responsibility is exercised through the user's own quality system and standard operating procedures and is not transferred to or shared with Certara by virtue of using the product.
No CoAuthor output — whether generated by an individual prompt or the Generate All function — should be submitted to a regulatory authority without prior human review, quality control, and sign-off in accordance with the customer's quality management system.
QC Requirement Every piece of AI-generated content must be reviewed and approved by a qualified human author before inclusion in any regulatory submission. CoAuthor facilitates drafting; it does not replace author judgment or quality oversight.
4. Transparency in the Product
CoAuthor surfaces AI disclosure at multiple points in the user experience to ensure that every user is aware when they are working with AI-generated content. These disclosure layers are in place across all AI-enabled surfaces:
Persistent AI indicator —an AI icon is displayed on all surfaces where AI functionality is active, providing continuous in-context visibility.
First-use notice — users are presented with an explicit disclosure the first time they interact with an AI feature within a session.
Onboarding email — all new CoAuthor users receive an onboarding communication that includes information about the AI capabilities in the product and the expectations for human review.
Welcome modal — a welcome screen presented to users upon first login includes AI disclosure information and directs users to this transparency statement. The persistent AI indicator displayed across CoAuthor AI surfaces is shown below: