Enterprise data platforms adapt for GenAI and agentic AI

Data is everywhere, and everyone uses it. Artificial intelligence 1.0 nudged organizations toward the right track. However, it was not until generative AI (artificial Intelligence 2.0) and agentic AI (AI 3.0) arrived in the industry that data became the star of the show.

Companies increased their hyper-focus on data infrastructure quality, accuracy, governance, security, ownership and sharing more than ever. With generative AI in play, organizations are realizing the importance of the state of their data platforms.

AI and machine learning developments have impacted how organizations perceive, consume, prioritize and protect their data. The entrance of generative AI applications in the business world has ushered in new demands on data quality. What used to be "close enough" regarding tolerance for raw analytics is no longer the case. Generative and agentic AI systems expose every irregularity in the data they consume, including typos and misinformation.

Data platforms are no longer behind the scenes. They're front and center.

With the fast-changing AI landscape, data quality has come to the forefront of every organizational strategy in data for several reasons:

• Data collection. Generative AI and agentic AI have raised the stakes for data collection. Organizations now need validation, classification and context of the data sources from which the information comes. This includes real-time data and data in motion. These "touchpoints" are becoming more evident as "data at the source" becomes a common practice. With edge computing, AI is being built on the perimeter of the enterprise, validating it where it's created before the system sends it upstream for further enrichment or utilization.
• Governance. Advances in AI have revolutionized governance concepts. They have become more agile in covering new processes and designs. In today's AI landscape, organizations are focused on data security and compliance as part of a broader effort to govern data infrastructure throughout its lifecycle. This approach goes beyond meeting compliance requirements to ensure what is needed today for AI systems to avoid biased or inaccurate results. AI governance must address the ethical dimensions of AI systems. This includes how and where they are being used, any inherent bias and the effects of that bias, and transparency around the origin of the data they are trained on.
• Data security. AI has changed how organizations think about the security and sovereignty of their data. As access to personal and corporate sensitive information expands, stakeholders are tightening their reins over where and how data is shared. Advanced pattern-matching algorithms allow AI to identify individuals in data sets that organizations have once pseudo-anonymized. This previously posed no issue before AI was adopted by many enterprises. Access controls, encryption, masking, key management and location-based data policies have also become the norm.
• Citizen data users. The way organizations use and engage with data has also dramatically changed. Visual engagement using generative AI tools allows non-data personnel to query the data through prompts using natural language. This advancement in data visualization to bring data closer to business users has resulted in non-data teams engaging with the data. However, this comes with some new constraints since they may not understand the limitations present in the data and its potential biases in natural language AI applications.

If an organization is looking to take advantage of the AI innovations that are currently taking over data centers, here are some quick considerations that can guide organizational thinking:

• Make data strategy core to the organization, from the board room to the data entry and data users. Everyone has a part to play in a data-driven organization.
• Invest in data platforms designed to support end-to-end data quality and governance. This includes data lakehouses, data products, data marketplaces, observability tools and governance frameworks.
• Centralize governance as much as possible. Security is hard to manage if it's fragmented. As organizations introduce AI systems, there is a big focus on AI governance. Are the AI outputs and actions by AI agents in line with the organization's strategy, ethics, accuracy and security policies?
• Expand data literacy beyond the data team. Include the entire organization for more critical understanding and skepticism toward AI-generated insights.

Data platforms are no longer behind the scenes. They're front and center, and it is up to the organization to decide which ones will win the data race when generative and agentic AI are fully unleashed on all corners of the organization.

Stephen Catanzano is a senior analyst at Enterprise Strategy Group, now part of Omdia, where he covers data management and analytics.

Enterprise Strategy Group is part of Omdia. Its analysts have business relationships with technology vendors.