AWS re:Invent 2025 - From Static to Dynamic: Fortive’s Next-Gen AI Platform on AWS (MAM356)

AWS re:Invent 2025 - From Static to Dynamic: Fortive's Next-Gen AI Platform on AWS

Introduction

• Presenters: Scott Warren, Americas Cloud Center of Excellence Lead at Capgemini, and Cedric Bordon, Global Technology Leader at AWS
• Overview of the "Fortive Brain" project, a next-generation AI platform built by Capgemini and AWS for Fortive, an industrial technology company

About Fortive

• Fortive is a large industrial technology company with 18,000 global employees operating in 50 countries
• Fortive's key focus areas include workplace safety, healthcare, and environmental safety, serving frontline workers, scientists, and patients
• Fortive operates several independent business units ("operating companies" or "OPCOs") under its umbrella

The Challenge: Fortive Brain

• Fortive's operating companies had disparate data sources, both structured (databases) and unstructured (SharePoint, Jira, etc.)
• Each OPCO had its own IT department and technology stack, making it difficult to standardize and govern a centralized AI platform
• Previous iterations of "Fortive Brain" were static, requiring manual updates whenever the underlying data sources changed

The Solution: Dynamic Fortive Brain on AWS

• Key goals:
  • Provide a centralized, governed, and secure platform for building chatbots and AI-powered applications
  • Enable dynamic integration with changing data sources without manual updates
  • Offer a user-friendly interface for OPCO teams to easily add new data sources and build new applications
• Architecture:
  • Web application hosted on AWS Fargate, accessed via CloudFront and WAF
  • Backend powered by API Gateway and AWS Lambda
  • Two-stage agent system:
    1. SQL Query Agent preprocesses requests and determines which data source to query
    2. Action Group Agent connects to the appropriate data source using the Model Context Protocol (MCP)
  • Structured data sources (e.g., Amazon RDS) integrated via MCP
  • Unstructured data (e.g., SharePoint) processed using Amazon Bedrock knowledge bases and OpenSearch
• Key features:
  • Seamless integration with various data sources (structured, unstructured, software engineering)
  • Dynamic data access, allowing real-time queries without manual updates
  • Standardized, user-friendly interface for OPCO teams to add new data sources and build new applications

Business Outcomes and Impact

• Rapid 8-week development of the initial proof-of-concept, demonstrating the platform's capabilities
• Enabled Fortive's operating companies to build chatbots and AI-powered applications without the need for extensive IT expertise or custom development
• Provided a scalable, performant, and secure platform for Fortive to expand its AI capabilities across the organization

Future Enhancements

• Integrate additional data source types, such as Oracle databases and Jira
• Leverage newer AWS services like Amazon Kendra to further enhance the platform's capabilities
• Utilize Amazon Kendra to streamline the development lifecycle, including automated database creation, MCP server setup, and testing

Conclusion

The Fortive Brain project showcases how Capgemini and AWS collaborated to build a dynamic, scalable, and user-friendly AI platform that enables Fortive's operating companies to leverage their disparate data sources and accelerate their AI-driven initiatives.