Drive has 700+ articles for digital transformation leaders written by StarCIO Digital Trailblazer, Isaac Sacolick. Learn more.
It may be time to finally fix the data management issues facing CIOs in large enterprises and SMBs.
Data silos, replicated data stores, separate architectures for data warehouses and data lakes, multiple data pipeline technologies, data security gaps, non-real-time materialized views, batch data integrations, databases with bloated stored procedures, yada, yada, yada – these are some of my favorite things.
Let’s call it data management debt.
There are some workarounds for data management debt issues. Just about every major SaaS provider offers a data fabric that, in theory, can help address data silo issues. Need to move data without disrupting its original source? It’s easier to build data pipelines today than ever before, and there are plenty of tools to help with data integration.
So, should CIOs kick the can down the road and focus on other forms of AI debt and AI cost debt?
Not so fast. There’s one cost that’s all too often overlooked – the time, complexity, and cost data engineers, data scientists, and business analysts waste dealing with data management complexities.
Data scientists and engineers have historically spent nearly 40% of their time on data prep and other data wrangling tasks. These include manual data operations, improving data quality, responding to data security issues, addressing operational incidents, complying with data privacy requirements, developing complex SQL queries, and so on.
Data engineers have nearly doubled the time they spend on AI over the past two years, from an average of 19% in 2023 to 37% in 2025, according to MIT. That number is likely much higher in 2026, so CIOs must identify data-wrangling productivity gains if data engineers are to be aligned with AI objectives.
Let’s also consider another set of issues. Data management debt is one reason business users may lose trust in primary data assets, slowing efforts to become a data-driven organization. On the other hand, ungoverned citizen data scientists creating dashboard after dashboard of calculations, predictions, and derived dimensions may deliver short-term value, but beneath the surface is a mess that needs unraveling.
CIOs need data to fuel AI innovation, address data management debt, and ensure data governance and security are in lockstep. Most importantly, data debt and all this data management debt can take a significant toll on a CIO’s AI strategy.
Addressing data management debt can be a digital transformation force multiplier by reducing data security risks, improving employee experiences, and enabling AI innovations. So I went to the Snowflake Summit to seek solutions for turning data management debt into an AI accelerator. Here’s what CIOs need to know.
Snowflake CoCo (formerly Cortex Code) is an AI development tool for “builders,” including data engineers, data scientists, and other data analysts looking to code solutions. According to Christian Kleinerman, EVP of Product at Snowflake, “The answer is CoCo” to much of the work builders need to code to support data pipelines, transformations, model development, data product building, and other data asset creation.
A second solution, Snowflake CoWork (formerly Snowflake Intelligence), is an assistant for all citizen data scientists, analysts, and business leaders (what Snowflake calls knowledge workers) who want to use data for daily decision-making.
Learn more about these two capabilities in my Driving Digital Standup video.
There are several tools for vibe coding and spec-driven development, but they are mainly geared toward building applications and AI agents. CoCo brings vibing to your organization’s data-driven builders.
And while just about all SaaS platforms have out-of-the-box AI agents and agent development tools, you’ll need to use their data fabrics and MCP servers when building workflows outside of their native capabilities. CoWork should be explored when knowledge workers just need to analyze and ask questions across data sets without friction, because the data is already centralized.
It’s fairly common to have separate data management platforms for structured and unstructured data – an issue Snowflake can address.
Even when centralizing unstructured data assets, many organizations purchase additional tools for search, content enrichment, and semantic analysis. Some CIOs may have yet another layer of tools for pulling intelligence from audio, video, and other media formats.
Snowflake showcased its multimodal AI capabilities for unstructured data, enabling CIOs to get more value from these sources.
Snowflake’s multimodal AI may be new for midmarket CIOs to explore without the complexity of procuring additional tools. For enterprise CIOs, multimodal AI capabilities may be a consolidation opportunity.
CIOs need to balance AI strategy with governance – not treat them as opposite sides of the same coin. Snowflake illustrated several governance and guardrail capabilities that CIOs can use in parallel to their AI innovation initiatives. Some examples
Getting value from AI is pushing CIOs to drive teams above the speed limit. Implementing governance and guardrails is playing catch-up in many organizations. AI is also introducing new data security risks. CIOs need data management solutions that support innovation while providing guardrails and governance capabilities to address real issues when scaling AI across the organization.
Many CIOs are stuck with too many AI POCs. Others deploy AI agents on back office functions where AI is reshaping the business but not transforming it yet. What’s the number one source of friction? People and change management, but what’s number two? Data trust, data debt, and data management complexities. Snowflake offers many capabilities under one platform, providing CIOs with opportunities for consolidation, simplification, innovation, and scalability.
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。