AWS re:Invent 2025 - Clinical Trials: AbbVie's Enterprise AI-Powered Document Creation Platform GAIA

AWS re:Invent 2025 - AbbVie's Enterprise AI-Powered Document Creation Platform GAIA

Overview

AbbVie, a leading pharmaceutical company, has developed an enterprise-wide AI-powered document creation platform called GAIA to transform their business processes, particularly in the area of clinical trials. This presentation by AbbVie and their partners Accenture and AWS provides insights into the journey, challenges, and future plans for this innovative platform.

Motivation and Vision

• AbbVie recognized the need to move beyond point solutions for automating individual documents and instead develop a scalable, adaptable platform that could handle the thousands of documents required across their clinical trial and R&D processes.
• The goal was to create a "Lego block" approach, with reusable components and AI-powered automation, to reduce manual effort, improve standardization, and increase the speed and quality of document creation.
• Partnering with AWS and Accenture was a strategic decision to leverage external expertise and accelerate the platform's development and deployment.

Key Features and Capabilities

• GAIA is a human-in-the-loop platform that can automate up to 90% of the manual effort required to create various regulatory and clinical documents, including Clinical Study Reports (CSRs), Periodic Safety Update Reports (PSURs), and New Drug Application Annual Reports.
• The platform ingests data from multiple sources, applies business logic and AI-powered transformations, and generates the first draft of documents, which are then reviewed and finalized by human subject matter experts.
• GAIA incorporates role-based access controls, in-browser rendering and downloading, a document orchestrator, an enterprise prompt library, and integrations with various data sources and language models.
• The architecture is designed to be scalable, adaptable, and future-proof, with the ability to incorporate emerging technologies like agents and context engineering.

Implementation Challenges and Lessons Learned

• Significant change management was required to educate business users on the capabilities and limitations of AI-powered document generation, as well as to establish a shared understanding of accuracy, writing style, and formatting requirements.
• Defining the value proposition and establishing appropriate metrics for success was an iterative process, as the team had to adjust their initial assumptions and baselines based on the actual automatable portions of each document.
• Building a cross-functional team with diverse skills and prior experience in AI and regulated industries was crucial to navigating the technical and organizational complexities of the project.
• Separating the business writing design from the technical design enabled the team to better capture the tacit knowledge of subject matter experts and translate it into an automated system.

Results and Future Plans

• By the end of 2023, GAIA will have automated 26 document types, saving an estimated 20,000 hours per year.
• AbbVie aims to expand the platform to automate over 350 documents and save more than 115,000 hours annually by 2030.
• The team is continuously enhancing GAIA's capabilities, including the integration of agents, context engineering, and self-learning mechanisms to further improve the quality and adaptability of the document generation process.
• The platform is being scaled across the enterprise, with plans to enable headless document generation and explore the use of domain-specific language models to power contextual automations.

Conclusion

AbbVie's GAIA platform represents a transformative approach to document creation in the highly regulated pharmaceutical industry. By leveraging AI and a scalable, adaptable platform, the company is able to streamline its clinical trial and R&D processes, improve quality and consistency, and free up valuable time for its subject matter experts to focus on more strategic initiatives. The lessons learned and best practices shared in this presentation can serve as a valuable guide for other organizations looking to embark on similar AI-driven transformation journeys.