Anand Ramamoorthy, Director APAC Data Governance and Quality at Informatica
Low quality of data is the biggest hindrance in the enterprise leap from pilot stage AI-systems to production ready AI agents, as per Anand Ramamoorthy, Director APAC Data Governance and Quality at Informatica from Salesforce in an exclusive conversation with businessline.
Following the company’s Data and AI Summit in Mumbai on June 11, Ramamoorthy said many enterprises voiced concern about the quality of their data with one company flagging 73 per cent of its data as “bad”.
“Data quality is the biggest impediment to translating these agents that you’re building to make it production-ready. It’s critical to have a data governance capability. The problem is the manual effort slows things down. So, even though it’s critical, they lose patience thinking that they’re not getting the value,” said Ramamoorthy.
Despite this hurdle, Ramamoorthy pointed out that traditional data management norms expect humans to be in the loop of workflows or in the interpreting of data. This means that the context when using data needs to be set by a human that can understand nuance rather than and AI agent that is liable to hallucinate.
“Data itself is not always incorrect. It is sometimes ambiguous, depending on the usage, the interpretation can be different. So, when you apply the agents on top of that, it amplifies that problem. That can lead to hallucination, the outcomes can be different,” he said.
To address these concerns, Ramamoorthy suggested three key pillars in the world of agentic AI: machine readable metadata with trusted context, AI-ready data products and AI-assisted data stewards that take up the task of providing governance policies around ownership, accountability, definitions, etc.
“It’s about how do we unlock the value of these data stewards who are good at functional and business-oriented conversations. How do we instill that into the manual effort rather than focus on mundane, repeatable manual tasks?” he said.
Published on June 16, 2026
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。