Only 22% of organizations surveyed by Gartner had successfully scaled artificial intelligence (AI) across multiple business units or adopted an AI-first approach, even as 85% of surveyed functional leaders planned to increase AI spending in 2026. That gap can become costly when tools enter the enterprise faster than departments can coordinate purchasing, prepare data, establish safeguards, and measure returns.
“Having a defined operating model and focusing on readiness before giving access to tools is key,” said Kevin Rogers, Practice Director of Data and Analytics at Trace3. “You don’t want tool adoption to happen faster than organizational readiness.”
Disconnected AI Projects Can Multiply Costs
Deloitte’s 2026 State of AI in the Enterprise report found that workforce access to sanctioned tools grew from fewer than 40% to around 60% in one year. Yet only 30% of surveyed organizations were redesigning key processes around AI, suggesting that broader access does not necessarily translate into changes in how work gets done.
When departments act independently, the cost can extend beyond multiple software subscriptions. Rogers described companies that purchased separate instances of the same type of cloud data platform, one for security vulnerability analysis and another for enterprise data and analytics, without recognizing the opportunity to consolidate their investments. “Two separate purchases, two separate environments, and no idea that they could have had economies of scale,” Rogers said.
The duplication can extend into the data work itself. Rogers said separate teams may each prepare the same enterprise data for their own AI use cases, repeating work across platforms and data pipelines.
A shared inventory of existing tools, planned pilots, and data resources can help organizations identify overlap before additional spending occurs. “There may be pockets of AI expertise throughout the company, which really should be tied together by a center of excellence,” Rogers said.
Prioritize AI by Value, Feasibility, and Actual Cost
Once organizations understand what they already have, the next step is deciding which AI use cases deserve further investment. Rogers recommends evaluating projects by business value and technical feasibility. High-value projects with manageable technical requirements can provide early evidence before companies commit more resources.
The financial analysis also needs to account for costs that may be overlooked during planning, including employee time spent on manual tasks, model usage, and ongoing operating expenses. “It’s important to measure the baseline, how much that process costs before AI, and how much it costs after implementing AI,” Rogers said.
That comparison becomes harder when companies adopt usage-based AI models. Unmonitored consumption can reduce or erase the savings an initiative was intended to produce. Rogers said organizations should track AI usage and establish guardrails around token consumption, particularly when relying on expensive models, so operating costs can be measured against the value each initiative delivers. “It’s all about showing results as soon as possible, getting a quick win, and earning the attention and investment to go after the next one on the list,” Rogers said.
Data Readiness Determines What AI Can Deliver
Preparing enterprise data for AI can expose problems that were easier to overlook in existing systems. Information may be scattered across platforms, inconsistently structured, or governed by access permissions that do not reflect how AI applications will use it.
Rogers said data readiness can remain difficult even as AI tools improve because data architecture and modeling require specialized expertise. Internal teams may also be maintaining existing systems at the same time they are expected to build new AI capabilities. Some organizations, he said, can spend years working through those foundational issues.
Trace3 works with organizations to assess their data environments, identify shared infrastructure, and support the technical work needed to prepare information for AI use. “The key is harnessing the data that’s proprietary or enterprise-specific, something competitors don’t have,” Rogers said. “You have to make sure that data is available, ready, centralized, governed, and secure, with enough context for AI systems to understand what it represents and how to use it.”
