The data management space has been transformed by artificial intelligence into one that not only seeks to prise maximum value from institutions’ data but also spends almost as much resource ensuring the quality of that information.
Without optimised data, the AI and agentic tools to which it is deployed will churn out sub-optimal results. It’s no surprise, then, that data quality is a recurring theme that threads through the programme of the 15th annual edition of A-Team Group’s Data Management Summit New York City.
A foundational session will be the panel discussion themed Scaling With Trust – Data Quality Engineering As A Driver Of Business Performance, which will examine best practices for alchemising data from a commodity to an asset that safeguards AI reliability while also achieving its long-established intent of bringing cost savings and revenue-generating opportunities.At the core of the conversation will be the recognition that the demand for, and even definition of, data quality has been fundamentally changed, said Sumanda Basu, head of data quality, data risks, controls and data domain lead at Société Générale.
“For years, data quality was viewed primarily as a discovery and remediation exercise, something organisations addressed after problems surfaced,” Basu, a panelist in the discussion, told Data Management Insight. “That mindset is no longer sustainable in an AI-driven world.
“As firms scale advanced analytics and generative AI, data quality engineering must move upstream and become an embedded part of the data lifecycle. The most successful organisations are adopting a ‘shift-left’ approach, designing quality controls directly into data products, pipelines, and business processes rather than relying on downstream cleansing efforts.”
Damaging Results
By feeding poor data into models, organisations run the risk of producing potentially damaging outputs that have the potential not only to lose them business but also run afoul of regulators. Data errors can also bring reputational damage.
Consequently, institutions are clamouring to set in place data quality assurance procedures to ensure their expensive new AI and agentic applications run smoothly and without issue.
Multiple reports have pointed out the difficulty capital markets institutions face as they seek optimal data for their AI models.
The same applies to private-market participants, who are similarly turning to AI and agentic automation to streamline operations in a space that has been traditionally lacking in complete and regular datasets. Indeed, sourcing good-quality data was cited as the leading challenge to private and alternative asset investors in a poll of delegates during a recent A-Team Group Data Management Insight webinar.
Brian Buzzelli, Director, Head of Data and Digital Transformation at Meradia, implied that there is a sense of urgency in the new approach to data management.
“The ultimate objective is not simply cleaner data – a mature data quality engineering capability creates reusable, trusted data products that reduce operational friction, increase automation and accelerate AI adoption,” he said.
“The strategic question is whether firms can turn the economics of better data — lower operating cost, reduced risk and faster innovation — into a self-funding engine for transformation,” Buzzelli told Data Management Insight.
Discipline Demand
Buzzelli, who will moderate the panel discussion, also stressed the need for good data quality as agentic AI and specialist data products proliferate.
“Data quality is moving from a control-oriented data management discipline to an engineering capability directly tied to operational performance, AI reliability, and business value at scale,” he said.
The panel will discuss a variety of topics relevant to real-world applications of data. Among them, speakers will look at:
- How firms are transitioning from reactive data cleansing to proactive ‘shift-left’ data quality engineering
- The unique data quality challenges presented by retrieval-augmented generation (RAG) architectures
- How machine learning can be deployed to autonomously detect anomalies and heal broken data pipelines
- The role of data observability in moving from static dashboards to continuous AI validation
- How firms manage data drift in real-time to ensure AI models remain accurate and deterministic
- How firms can reframe data quality metrics in terms of operational efficiencies or savings to help fund AI, or drive business performance
AI-Ready Data
An oft-repeated mantra within the modern data management ecosystem is the need for organisations to ensure their data is AI-ready. But another panelist, Mick Hittesdorf, cloud product architect at KX, points out that this is a phrase loaded with meaning but often misinterpreted. He echoed a comment from Buzzelli, that data quality is an essential prerequisite of trustworthiness when it comes to feeding AI models.
“Proving your market data is AI-Ready – so you can truly trust your data and the outputs that AI-enabled workflows produce from it – is one of the most urgent drivers of data quality engineering today,” Hittesdorf told Data Management Insight. At the panel, he will stress that “a systematic, metrics-driven, data quality programme is an essential prerequisite for earning and keeping the trust of your data consumers”.
The panel will also include Michael McCarthy, vice president – enterprise data management at MFS Investment Management; Naveen Chavali, director – product solutions at NeoXam Americas, NeoXam; and Lydia McCurdy, Informatica account solution engineer at Salesforce.
- Data Management Summit New York City will be held on September 17 at @Ease, 7th Floor, 605 Third Avenue, New York. Attendance can be reserved by clicking here.
Subscribe to our newsletter
