Enterprise AI strategy surges in importance for IT buyers

Enterprise AI strategy: Improve decision-making, productivity

Drilling down into AI's heightened enterprise significance reveals a specific focus on GenAI. Survey respondents identified it as the No. 1 data science and AI/ML technology in which organizations planned substantial investments for the coming 12 months.

GenAI led by a considerable margin: 53% of respondents cited it, while large language models (LLMs), which underpin GenAI tools, ranked second at 35%. Enterprise ML platforms and ML operations platforms placed third and fourth, with 32% and 31% of respondents selecting them, respectively.

The investment priority status for GenAI and LLMs reflects a shift in how businesses apply those technologies and the benefits they realize as a result. Many organizations initially focused on content creation -- drafting email messages and creating marketing materials, for instance. Those applications haven't gone away, but a more recent pattern has businesses using the latest crop of GenAI tools to bolster their core processes.

Here, the goal is to boost business results through improved decision-making, cost reduction and increased productivity. The following examples help illustrate the trend.

Kroger unit improves 'judgment decisions'

"AI has opened a whole new opportunity for us as a business. It's around being able to use data to improve decision quality across the board," said Todd James, chief data and technology officer at 84.51°, grocery chain Kroger's retail data science subsidiary. The unit is named for the longitudinal location of Cincinnati, where Kroger is headquartered.

AI improves what James calls "judgment decisions," in which, historically, a human would take an action based on corporate training, policies and procedures. AI and ML, however, extract insights from data to help drive stronger outcomes, he added.

AI has opened a whole new opportunity for us as a business.

"The benefit is clear," James said. "Using data, activated through machine prediction, is a relatively new lever that can be used -- alongside training and procedures -- to better inform decisions."

In one example, the subsidiary created an app that uses AI-enabled dynamic batching to reduce the distance employees walk in Kroger's grocery stores to find and pack items for customer pickup orders. Employees using the app, which also includes a digital map of stores, have cut their shopping trips by 10%, according to James. This, in turn, has reduced the lead time for pickup orders to less than two hours.

In addition, his team repurposed the store app's routing algorithm to optimize the routes Kroger delivery trucks take from distribution centers to stores. An initial test showed an average mileage reduction of 8.62%, James noted.

Discover taps GenAI as decision 'safety net'

Discover, the credit card and financial services company, uses generative AI to process customer call transcripts. That application taps GPU-supported AWS instances, Nvidia drivers and Llama 2 LLMs.

The objective is to make sure the company's agents contact customers according to their preferences. Customers might, for instance, ask to be contacted through one communication channel, on one phone number or at a certain time of day. Adhering to those preferences improves customer experience, a plus for any business. But in the financial services industry, respecting contact preferences is also a regulatory requirement, Discover CIO Jason Strle noted.

"Following certain business processes is an important part of how we manage risk and how we are expected to manage risk," he said.

In this case, the risk is that an agent fails to fully grasp a customer's wishes and chooses the wrong contact approach. Discover's GenAI application, however, supports agents "at the decision point," according to Strle.

The application checks call transcripts "to make sure that we have not missed any place where the customer may have communicated their intent," Strle said. "It's a safety net. The generative AI solution is an additional validation step to ensure the human executes the process correctly."

Strle said the financial services industry sees both opportunity and risk in generative AI. The transcript processing use case shows how the technology can provide business value and reduce risk exposure at the same time, he noted.

"What I'm really looking for is combining the two and saying, 'If we're concerned about the risk created by GenAI, let's use GenAI to manage the risk,'" he said.

Liberty Mutual sees 'step change' in GenAI

Liberty Mutual Insurance has been using AI and ML for years in risk prediction efforts. The arrival of generative AI, however, signaled a transition that raised the technology's profile at the Boston-based insurer.

"When GenAI came, there's no question about it: It's a very exciting step change," noted Tony Marron, managing director of Liberty IT, Liberty Mutual's technology arm, which is headquartered in Belfast, Northern Ireland, and also has two offices in Ireland.

The insurance company has deployed a nonpublic version of ChatGPT, which it calls LibertyGPT. About 25% of Liberty Mutual's 40,000 employees use the tool. The insurance company has 16 GenAI use cases in production and another 33 in research and development. At scale, those initiatives will have an estimated value of more than $100 million, based on projected productivity gains and cost savings, according to Marron.

But the benefits aren't entirely financial. Marron also cited the importance of improved data accuracy and employee experience.

"We are finding value can come in other forms," he said. "Those are harder to put a dollar figure on, but that doesn't mean we ignore them."

The insurer's GenAI use cases currently focus on augmenting employee activities -- what Marron called "copilot" applications. The next step, he said, will be to "bake" GenAI into the company's core systems and workflows.

"I think that's really exciting because the more information and the more context you are able to provide GenAI, the better the insights and the better the predictions," Marron said.

This embedded-GenAI phase could emerge in 2025 or perhaps later, because the task will be much bigger than creating copilots, he noted. But the upcoming crop of use cases, while more closely tied to business processes, will not be completely autonomous. Marron said the company will use a human-in-the-loop approach that lets employees overrule AI-based decisions.

Cybersecurity investment plans

Enterprise AI's higher strategic profile dovetails with plans for more GenAI spending. Similarly, cybersecurity's top-technology status translates into increased enterprise investment. As mentioned previously, the survey identified strengthening cybersecurity tools and processes as the No. 1 IT initiative for 2025, with 27% of respondents ticking that box as their top choice. Using AI and ML was a distant second on the list, at 16%. In addition, 72% of respondents said their organization would increase spending on cybersecurity this year, ranking it first on that list by a wide margin.

In network security, firewalls and zero-trust network access (ZTNA) led the way, with 39% of respondents citing the former and 38% the latter for significant investments. Those technologies also topped the list of network security technologies in the 2024 survey but with ZTNA narrowly leading firewalls then.

A similar pattern emerged in market research and consulting firm Gartner's cybersecurity spending forecast, which has both ZTNA and firewalls growing in 2025. "Firewalls are going up, but it's ZTNA that is sort of taking over," said John-David Lovelock, research vice president at Gartner. "CIOs are migrating fairly quickly to zero-trust networks."

According to Lovelock, the cybersecurity sector is expected to expand at a 13.3% annual clip overall.

Generative AI, meanwhile, is growing in importance as a cybersecurity component, according to the Informa TechTarget survey. GenAI-powered offerings tied for 10th among network security technologies in the 2025 spending intentions survey, with 20% of respondents identifying them for planned investments. GenAI-based network security offerings ranked 16th -- and last -- in the 2024 survey, at just 6%.

Sizable percentages of respondents also said their organization planned to invest in GenAI-powered technologies for other facets of cybersecurity. Those figures were 25% each for cloud security, SecOps and application security, plus 20% for identity-related security technologies.

Many buyers will first encounter GenAI in a security setting when it's embedded in vendor offerings.

"In 2025, the main way organizations will start to leverage generative AI in cyberdefense will be as their vendors start to introduce it into their products," said Richard Watson, global and Asia-Pacific cybersecurity consulting leader at advisory and professional services firm EY.

The top product categories featuring GenAI will include threat detection and response, identity and access management, and testing and vulnerability management, he said.

Increasing IT complexity

Investments in nascent technologies and cybersecurity aim to improve efficiency, buttress business processes and boost resilience. But they also contribute to IT complexity. In the Informa TechTarget survey, 60% of respondents said their IT environments had become more complex over the past two years. Nineteen percent said the level of complexity hadn't changed and 15% said it had declined, while 7% were unsure.

Technology and business leaders cited the expanding and changing cybersecurity landscape as the top reason for rising complexity, with 42% of respondents pointing to that consideration. They rated the need to incorporate new and emerging technologies as the No. 2 complexity factor, chosen by 36%.

TEKsystems survey results, published in January, also highlighted the intricacy of IT as an issue. The company's "2025 State of Digital Transformation" report said the complexity of current IT environments, coupled with "siloed mindsets," were the biggest obstacles in the path of digital transformation. One-third of the 855 IT and business leaders who were polled late last year cited those issues, according to TEKsystems, a technology and business services provider based in Hanover, Md.

A push among organizations to pursue smaller digital transformation projects of shorter duration could help address the complexity challenge. The portion of digital transformation projects valued at $10 million or more dropped to 22% in the latest TEKsystems survey, down from 30% the year before. At the same time, the survey showed a corresponding increase in the portion of projects valued between $1 million and $4.9 million: Those rose to 35% of project portfolios, up from 27%.

"The size of the engagements as well as the project lifecycles are shrinking," TEKsystems CTO Ram Palaniappan said. The smaller, quicker projects speed up ROI for board-level executives who don't want to wait two years for the results of a big-bang IT project, he noted.

The nimbler approach to digital transformation also accommodates the complexities of dealing with fast-paced technology changes in fields like AI. Palaniappan cited the sudden arrival of ChatGPT in 2022 and more recent developments such as DeepSeek. Smaller projects and a structured program for conducting pilot deployments let organizations prove out new technologies and add incremental features over time, he added.

Enterprise Strategy Group made a similar point in its report on the Informa TechTarget spending intentions survey. The report said integrating AI technologies with existing systems and applications "inevitably complicates environments." But it added that there's "an unmistakable fear of missing out" on AI capabilities and predicted that confidence in GenAI's potential business value from new trials and initiatives "will sustain increased funding for the foreseeable future."