With so many data management platform categories and platforms, it can be challenging to determine what to review and prototype. Sitting on legacy data platforms and outdated practices can drag down the CIO’s AI agenda.

Looking over Gartner’s list of Magic Quadrants, at least twenty have data management capabilities. Matt Turck’s Machine Learning AI & Data Landscape has over 2,000 logos and is truly maddening. No wonder many CIOs have analysis paralysis on data platforms.

The Magic Quadrants and Turck’s landscape don’t account for all the platforms storing enterprise data. For a frightening picture, consider the 2025 marketing technology landscape by chiefmartec Scott Brinkler, which showcases over 15,000 logos. Yikes!

In this post, I asked ten questions about data management, quality, security, and pipelines. Experts weigh in on the importance of identifying solutions, and I share my perspectives.

1. How are you monitoring your data pipelines?

Edward Calvesbert, VP of product management at IBM watsonx.data says, “Manual data observability leads to major preventable risks like harmful model outputs, compliance violations, and poor business decisions, necessitating automated metadata collection and prescriptive anomaly thresholds.”

My take: There are data integration, data streaming, and data pipeline platforms that provide both dev and ops capabilities. However, IT often focuses on developing integrations and leaves off the important ops work to monitor pipeline reliability and performance. Evaluate dataops observability and develop operational KPIs to increase IT ops’ attention on continuously improving pipeline reliability.

2. How are you developing trust in data?

“CIOs must take charge of data governance by allocating resources to continuous quality monitoring and automated tools,” says Paul Boynton, co-founder and COO of Company Search Incorporated. “Otherwise, they are vulnerable to stale analytics, eroding confidence in dashboards and creating operational inefficiencies.”

My take: Look to develop a data health metric focusing on data quality. Define KPIs around dataops, data security, and data governance. Then, develop metrics that measure data’s timeliness and age out stale data from ML and AI models.

3. How can you automate web data extraction when you need it?

“CIOs should be reviewing data extraction platforms for their ability to deliver structured, compliant data in real-time,” says Luis Lacambra, head of product at SOAX. “Automated parsing and deduplication at the point of collection is now a must-have to prevent AI models from being trained on noise. Platforms with adaptive unblocker infrastructure, like web data API that ensure access to target data from websites with advanced defense systems or with dynamic content are also critical for reliable data extraction.”

My take: CIOs should make sure that architects are well-versed in the right tools for the data integration job. More importantly, they must understand compliance when integrating third-party data sources.

4. Where is your data coming from?

“When reviewing data platforms for GenAI, CIOs should prioritize systems with strong lineage and metadata management,” says Chris Hendrich, associate CTO of application modernization at SADA. “The value here is ensuring data traceability, while the technical capability should include automated data cataloging that integrates with GenAI workflows. This foundation breaks down data silos, making trusted data shareable for accurate, context-aware GenAI responses.”

My take: CIOs in regulated industries understand data lineage’s importance from a compliance perspective. The capability is even more important today when using multiple data sources to feed AI models and agents.

5. How are you centralizing access to data for AI applications?

Calvesbert of IBM watsonx.data says, “Without the ability to combine structured and unstructured data through a modern data fabric, expensive AI agent initiatives will fall short of expectations and return on investment, making insufficient data accuracy and governance a critical concern as organizations deploy agentic AI.”

My take: Organizations that are multicloud and suffer from SaaS sprawl would benefit from using a data fabric.

6. How are you automating data governance to support AI?

 “Soon, agentic AI will change the data governance landscape, and AI-driven systems will request, analyze, and act on data at a speed and scale that human-driven processes can’t match,” says Steve Touw, CTO of Immuta. “In order to keep up with this surge, CIOs must evolve their data governance strategies with dynamic, policy-driven automation that removes toil and scales enforcement at machine speed. Otherwise, organizations will hit a breaking point, increasing the risk of security breaches and non-compliance.”

My take: See my previous articles on developing a balanced AI governance strategy and six AI/data governance non-negotiables. Review platforms and partners that help automate some of the work. But, it’s the CIO and CDO’s responsibility to identify data owners and establish proactive data governance plans.

7. How are your AI teams evaluating data biases?

 “The rush to adopt AI introduces risks from inaccuracy and bias, leading to flawed outcomes that can impact or even injure customers,” says Dan  Spurling, SVP of product management at Terradata.  “Some AI initiatives can and should move rapidly, but some must be anchored in trust. Trusted AI requires a foundation of reliable data, governance, oversight, and transparency into decisions – all delivering AI outcomes that drive the business forward in today’s competitive markets.”

My take: See my review of Winning with Data Science, which has an excellent chapter on data bias. We discussed data privacy at a recent Coffee With Digital Trailblazers. Read or listen ourrecommendations for CIOs on taking control of data.

8. How are devops teams testing AI-generated code?

“CIOs must approach AI-driven application development and code generation with strategic caution, not fear – anchored by a governance model that ensures both accountability and resilience,” says Deepali Bhoite, CISO at Quickbase. “While creative trends like vibe coding and slopsquatting reflect AI’s expanding role in innovation, they also signal the need for structure. In some circles, testing AI-generated code may seem contrary to the ethos of frictionless development, but few organizations are truly prepared to operationalize these tools at scale without introducing exposure.”

My take: My article on AI in code generation has plenty of stats citing benefits. But my bet is on how low-code platforms with AI agents for building applications will win the productivity race. Why is this important for data management? Think about the importance of accelerating application modernization to ensure key workflows are ready for the genAI era.

9. How can your AI tools obey data security controls?

 “When reviewing platforms, CIOs must ask how seamlessly their data platform integrates with their AI platform of choice,” says Hendrich of SADA. “The strategic value is eliminating the security gaps and development friction that arise from bolting two systems together. Look for the technical capability where security controls from the data layer are natively understood and enforced by the AI layer out-of-the-box.”

My Take: Hendrich asks a good question, one that I haven’t seen a simple solution for CIOs to rely on. Leave me suggestions in the comments.

10. How are you monitoring and tracking AI agents?

“As AI agents proliferate in organizations and take on more “human” responsibilities, we are going to need a new, reimagined version of HR for our AI,” says Jed Dougherty, head of platform strategy at Dataiku. “Enterprises will require an agent management system that oversees the quality, cost, safety, and organizational structure of agents that are moving from an advisory role to actually performing critical tasks and making business decisions.”

My Take: The gold rush to experiment with AI agents is leaving many operational questions behind. It feels like we’ll be griping about AI agent debt and shadow AI agents soon. AI agents gone rogue will be scary.

What’s your take on the future of data management?