End-to-end lineage that enables robust data traceability is now considered the “heartbeat of an enterprise” and no longer a niche interest of data managers, according to an A-Team LIVE webinar.
Focusing on the importance of metadata to two particular use cases – regulatory compliance and artificial intelligence readiness – panellists agreed that without a solid lineage foundation, financial institutions could not hope to govern their data at all.Being able to trace the transformation of data from its ingestion into institutions’ systems to its ultimate end application was critical for both uses, especially for remediation and governance, and was critical to ensuring data is trustworthy.
The speakers comprised Murali Duvapu, Data Governance Executive at Scotiabank; Cheryl Benoit, Executive Director for Operational Risk at Mizuho; and Christian Bremeau, Chief Executive of Meta Integration, the event’s sponsor.
A-Team Group Data Management Insight Editor Mark McCord moderated the webinar, which was entitled End-to-End Lineage for Financial Services: The Missing Link for Both Compliance and AI Readiness.
Ensuring Trust
For compliance teams, end-to-end lineage provides the means to effectively meet regulatory obligations by providing certainty that the data disclosed to regulators is correct and accurate, and will forfend against potentially costly breaches.
That’s all the more important as regulators are demanding proof that reported figures can be shown back to the book of records. Furthermore, compliance often requires historical context, necessitating a metadata management platform capable of providing version- and time-management to compare data structures from previous quarters or years.
Regulatory pressure is also forcing the issue, the webinar heard. The EU AI Act, for instance, places responsibility for transparency not solely on data producers but on the consumers who utilise AI sources for end-user reporting. This necessitates comprehensive data lineage to provide a clear footprint of all sources used.
While lineage can manage the input and govern the output of AI, panellists cautioned that the decision-making process within AI models will never offer the same proven lineage of a decision as traditional reporting. Therefore, maintaining human oversight remains critical to success.
AI-Readiness
Data lineage is indispensable for leveraging AI effectively, moving beyond the simple “garbage in, garbage out” risk, the webinar heard. It assures a high level of data quality, which is critical for preventing AI models from generating incorrect or potentially dangerous outputs.
The panellists’ views were supported by polls of the webinar’s audience, which found that regulatory compliance and data quality improvements were the two applications that were benefiting most from the presence of robust end-to-end lineage.
Transitioning to such a state, however, is beset by challenges, particularly in terms of gaining enterprise-wide buy-in of such an objective. To overcome this cultural hurdle, panellists suggested focusing on automation of the process and on greater collaboration between companies’ business and technical users.
Subscribe to our newsletter
