In order to streamline the use of data in response to regulations within its industry, a global asset and wealth management firm partnered with Capgemini to introduce an AI-powered solution. This has accelerated the company’s ability to take action in response to regulatory issues by cutting manual effort related to governance and documentation by 50-70% while accelerating related workflows by 30-50%.

Taking the detective work out of data lineage

Because it operates in a heavily regulated industry and manages trillions in assets, a global asset and wealth management firm must respond to regulators within strict timelines and with proper evidence. So, knowing exactly where its data comes from and how it moves is critical.

But over decades of operation, the firm’s systems went through several waves of modernization, from mainframe and DB2 environments to on-premises platforms, a data analytics environment, and finally to the AWS cloud. Each wave added layers of complexity, and tracing how data reached its destination meant analyzing the code by hand.

Teams would study code bases and record the results in spreadsheets to extract data lineage manually. This process, running from extraction to data steward approval, could take more than six months, which led to slower regulatory responses and complaint resolution, putting both compliance and customer experience at stake.

To address this challenge, the company turned to Capgemini with the ambitious goal of using AI to make data lineage available on demand.

From proof of concept to enterprise product

The partners worked on a proof of concept (PoC) to launch the project. The company wanted to apply GPT models and retrieval-augmented generation to extract data lineage directly from PySpark code bases. In addition, the project team used context engineering techniques to assemble and enrich information from source code, metadata, business rules, and system dependencies. As a result, AI models gained the ability to generate accurate lineage, documentation, and governance insights.

Following a successful first application, the teams began scaling the platform feature by feature, with each taking roughly two to three months to develop. The project team then added automated documentation and business rule extraction before introducing entity-relationship (ER) diagram generation, and a data flow visualization that demonstrated how applications exchange information across the organization. Further repositories were incorporated with the scope later expanding beyond data pipelines to include digital applications and both Java and UI code bases.

As the platform expanded, AI capabilities evolved beyond lineage extraction to create business-friendly interpretations of technical assets. By translating code, transformations, and application logic into understandable business context, the solution established a foundation for future semantic-layer capabilities that can make enterprise data easier for both business users and AI systems to access and understand.

The platform has been scaled over the course of six months to connect lineage across different code bases enterprise wide and transform governance. Today, anyone can point the solution at a code base and request its lineage, after which a data steward can review and validate the results the next day. Finally, AWS provides the cloud foundation, with the firm’s data lake already running on the platform.

Fast and fastidious: Quicker answers, stronger compliance, and a $20 million outlook

The transformation cut months of manual work and delivered real results. Time spent on governance and documentation fell by 50–70%, and governance workflows accelerated by 30–50%. Meanwhile, approval cycles that once took six months or more are now completed in days.

We enabled the company to avoid fines by responding rapidly to regulatory issues. In addition, the project improved customer experience by enabling teams to trace the flow of customer data and rectify issues quickly when dealing with complaints.

The platform is now being scaled enterprise-wide, with around $20 million in savings projected over the next three years. In addition, there are further plans to extend the Gen AI solution across personal investing, institutional investment, and financial advisory services. Beyond improving governance today, the platform helps establish an AI-ready enterprise data ecosystem built on trusted lineage, explainable data foundations, and richer semantic understanding of business information.

Following the project, the partnership continues to grow by developing and scaling innovative solutions with Capgemini’s AI center of excellence, helping the firm stay agile, insight-driven, and focused on investor outcomes.