How Generative AI Transformed FaceUp's Secure Whistleblowing Platform

FaceUp is a secure and anonymous SaaS platform for whistleblowing. It is designed to help organizations track, review, and address issues like bullying and harassment in compliance with regulatory requirements. With a full suite of features for confidentiality, real-time reporting, and streamlined case management, FaceUp helps its customers ensure transparency and accountability. Whether incidents are reported by co-workers, students, or constituents, the FaceUp platform gives customers tools to build a culture of ethical behavior, respect, and integrity.

‍

The Challenge

As usage scaled, the platform users encountered technical limitations in managing reports efficiently, resulting in delays, workflow bottlenecks, and inconsistent outcomes for customers. Traditional whistleblowing systems require administrators to manually process and analyze large volumes of reports. The anonymity of whistleblowers makes follow-ups difficult, leading to incomplete information and slower resolutions. For example, potential whistleblowers often do not know which evidence is (or is not) required for the organization to see that the reported problem is dealt with appropriately. This lack of automated investigation and analysis capabilities creates challenges in pursuing potential wrongdoing, which in turn undermines operational efficiency and scalability for both the organization and its members. In other words, the demands of thoroughness and privacy obligations conflict with operational effectiveness.

‍

The Solution

To tackle these challenges, FaceUp partnered with CloudGeometry to leverage AWS-powered generative AI to create a more effective SaaS platform for its customers. The platform integrated advanced AI capabilities such as report summarization, dynamic follow-up question generation, and actionable recommendations for administrators. The solution utilized Retrieval-Augmented Generation (RAG) architecture, combining document retrieval with AI-generated insights to ensure accuracy and relevance. Key AWS technologies like Bedrock for hosting the Claude 3 Haiku model, OpenSearch for vector storage, and ECS for scalable application deployment underpinned the system.

‍

The Benefits

The Generative AI-enabled platform delivered transformative results:

• Efficiency Gains: Automating repetitive tasks significantly reduced claim processing times, enabling administrators to focus on high-priority cases.
  

• Enhanced Scalability: The solution seamlessly handled up to 1,500 concurrent reports without performance degradation or a meaningful increase in system costs.

• Improved User Experience:: AI-generated insights ensured faster, more precise decision-making while maintaining anonymity and security.

‍

Key Features

• Data Storage: Amazon S3 is used for scalable and secure data management, integrated with AWS Lambda and ECS, which enable real-time processing and auto-scaling capabilities.
  
  
• Generative AI: Claude 3 Haiku model hosted on AWS Bedrock for report analysis and recommendations.
  
  
• RAG Architecture: Combined vector storage with AI to enhance response accuracy and relevance.
  
  
• Cost Efficiency: Combination of open source models with well-orchestrated events-driven architecture allowed to bring down the processing cost to as low as less than 1 USD per report.

Conclusion

By integrating advanced Generative AI capabilities with AWS technologies, FaceUp transformed its whistleblowing platform into a more efficient and scalable solution that maintains the highest standards of security and confidentiality. The innovative use of RAG architecture and the Claude 3 Haiku model enabled faster processing times and more thorough report handling while keeping costs remarkably low. The successful implementation demonstrates how AI can enhance sensitive operations without compromising privacy, setting a new standard for ethical reporting platforms. FaceUp's experience shows that with the right architecture and technology choices, organizations can successfully balance operational efficiency with privacy obligations.