On Track to the Future: AI in Compliance Auditing
In a groundbreaking collaboration, Capgemini and Network Rail have embarked on a proof of concept to test the effectiveness of an AI-based call auditing solution. This innovative approach aims to determine whether communications adhere to the Safety-Critical Communication (SCC) compliance standards.
Client Challenge
Network Rail faced the challenge of conducting a feasibility experiment to develop an AI-driven solution capable of auditing calls for compliance with SCC standards. With a responsibility for billions of passenger journeys and freight annually, ensuring safety in communication is paramount.
Solution Overview
In partnership with Capgemini’s Applied Innovation Exchange (AIE), Network Rail utilized a proof-of-concept AI auditing solution throughout a designated test period. This collaboration was focused on assessing the solution’s ability to effectively audit calls, thereby improving the quality of communications vital for safety.
The Role of Network Rail
As a key player in the UK transport sector, Network Rail maintains over 20,000 miles of railway infrastructure, ensuring the safety and efficiency of operations. With thousands of engineers engaged in daily tasks, the organization manages approximately 5,000 phone calls daily, all of which must comply with strict SCC protocols.
Implementing AI to Analyze Calls
To explore the viability of an AI auditing solution, Network Rail engaged Capgemini to leverage their industry expertise and technical knowledge. This collaboration initiated a comprehensive review of the program’s objectives and available technologies, leading to a structured roadmap for implementation.
During this process, Network Rail provided 200 call recording files to train AI models, along with their SCC manual to facilitate conversation mapping. The project team developed a conversation mapping tool, incorporating a custom speech-to-text (STT) recognition model and a Natural Language Processing (NLP) algorithm. This model analyzes calls based on four key parameters: Clarity, Completeness, Compliance, and Focus.
Findings from the Feasibility Study
At the conclusion of the feasibility study, a detailed report was presented, which included findings, recommendations, and best practices for utilizing AI in auditing safety-critical communications. The tool was subsequently employed over a three-month period to regularly analyze recorded calls.
Insights and Recommendations
Through extensive technical analysis, Network Rail and Capgemini concluded that the NLP models performed effectively in identifying non-compliant calls and detecting missing information. Despite challenges such as background noise affecting clarity, the AI solution excelled at determining the portion of calls dedicated to SCC.
The results indicated that an AI-driven solution has significant potential for enhancing compliance assessments and improving operational efficiency. This experiment also provided guidelines for the ethical use of AI in analyzing safety-critical communications.
Future Prospects
With the insights gained, Network Rail is now equipped to scale the proof of concept for enterprise-wide deployment. The integration of AI technology will not only ensure communication compliance but will also lead to increased passenger safety and operational efficiency.
As organizations continue to explore the capabilities of AI in compliance auditing, the partnership between Capgemini and Network Rail serves as a leading example of innovation driving safety in the transportation sector.
