The 3 ‘no regrets’ AI moves for IT organizations

In my work with enterprise technology leaders, I’ve seen so many companies lay claim to AI rapidly (through pilots, tests and cross-functional experiments) that the enterprise AI moment increasingly feels like a gold rush. If it is, then leaders should take a lesson from history, from the real Gold Rush. During that time, the most durable wealth didn’t go to just any prospector. It didn’t even go mainly to the ones who struck gold. It went to the ones who built the railroads (and the equipment, the supply chains and the financial systems). It went to the builders of infrastructure.

In 2026, the year of scale or fail in enterprise AI, CIOs face a similar inflection point. The question is no longer where they should dig. It’s what infrastructure they should build. When I evaluate an AI-related investment, I come back to three questions:

• Will it show measurable value within 12 months?
• Will it build a durable enterprise AI capability rather than another pilot?
• Will it increase organizational capacity?

These are the value levers that matter. AI experimentation may generate headlines, but only infrastructure generates enterprise impact. That is why I believe CIOs should focus on three no-regrets moves: make knowledge a living enterprise asset, transform IT service management to deliver better outcomes and accelerate the software development lifecycle.

1. Make knowledge a living enterprise asset

Critical institutional knowledge exists everywhere: in ticket histories, engineering documentation, email threads, chat conversations and SharePoint repositories. Yet in most organizations, this knowledge is fragmented. It’s static and disconnected from the flow of IT work. And these conditions erode productivity. They leave engineers to spend too much time searching, service agents to escalate issues already solved and teams to rework solutions rather than build on them.

AI changes this dynamic. It allows companies to mobilize their documentation as never before, to turn static archives into dynamic knowledge as source data is generated automatically, refreshed continuously and embedded in workflows. With this turn, content becomes contextual and role-based, and it arrives at the point of need, saving every engineer (maybe every employee) 30 minutes that were lost daily to needless efforts. The impact of these gains scales quickly and often translates to millions of dollars annually.

We have seen this evolution play out with an industrial automation company whose knowledge was scattered across so many platforms that human agents spent more time searching than resolving. But once they started using AI to generate, structure and embed knowledge in their service workflows, the company reduced handling time to the tune of $3 million annually. Just as importantly, by mobilizing all this knowledge, the company effectively laid a foundation for the governance that will support future AI agents.

When knowledge becomes structured, governed and trusted, it becomes fuel for everything that follows, and it improves the context by which AI agents can reliably scale. In this way, it enables agentic support, enterprise copilots and intelligent orchestration. In short, knowledge is more than a side project. It is infrastructure itself.

2. Transform ITSM from workflow optimization to outcome optimization

Across industries, service desks face a common challenge: they must handle growing ticket volumes and expectations. Yet they often must keep their headcount the same. They can mitigate some of these pressures through traditional ITSM platforms, which help to optimize their workflows. But mitigation is about as far as it goes.

The real solution lies in supplementing their ITSM with AI, which can unlock the kind of nonlinear support that changes the very economics of their work. Typically, this supplementing comes about in stages, each introducing a new category of ITSM-related, AI-powered use-cases: intelligent intake and classification, automated routing, embedded knowledge retrieval, guided self-service and ultimately, agentic resolution of repeatable incidents. What’s important is that, together or separately, all these stages reflect the same pull, not just toward ticket deflection, but toward the automated resolving of repeatable work.

We saw a similar pattern at a SaaS provider whose service desk faced rising ticket volumes and escalating support costs. By implementing AI-driven intake, automated triage and self-service capabilities, the organization not only deflected 43% of tickets in year one but also reduced handling time by automating triage interactions. Ultimately, the organization saved $6 million over two years while increasing support capacity. And they did it all without adding headcount.

Anywhere demand is predictable, AI can accelerate outcomes. It can cut resolution times, improve SLA performance and diminish tasks that are manual, repetitive, or both, saving organizational resources while improving service. And just as importantly, it can turn ITSM into a launchpad for enterprise AI. The capabilities developed here (intelligent orchestration, governed knowledge use, agent workflows) extend naturally into HR, Finance, Procurement and other shared services. And this insight implies that AI-related value can not only be proven in IT, but scaled.

3. Increase enterprise capacity through AI-enabled SDLC

Just as ticket demand is not slowing down, neither is digital demand. Backlogs are swelling, and for the most part, organizations can’t hire their way out of the problem. But they can use AI to accelerate their SDLCs. Within the SDLC, AI can generate and refactor code, create test cases automatically, detect defects earlier, optimize release pipelines and provide real-time engineering insights. As a result, it accelerates throughput, improves quality and enhances predictability, and all with few changes to headcount.

One of the clearest examples I’ve seen came from an energy company that gave AI-enabled development tools to a cohort of engineers, who used the tools mainly to generate and review code and modernize their workflows. They doubled their output. And what’s more, they reduced their revert rates by 79% and saved 4,600 annual hours. AI didn’t just improve their productivity incrementally; it unlocked step-change throughput and quality gains while expanding the organization’s capacity to deliver.

When adopted the right way, AI-enabled SDLC does more than accelerate the writing of code. It helps the enterprise to scale its digital capabilities. In an environment where demand outpaces capacity, this means IT organizations can absorb growth, innovate faster and expand delivery without adding headcount.

From AI curiosity to AI capacity

AI has changed the mandate of CIOs. They can no longer just experiment or declare success on the value of isolated use-cases. They must produce value across the enterprise, improving economics, expanding capacity and creating durable competitive advantage. An ideal way to pull these value levers are the three no-regret moves:

• Make knowledge the bedrock on which we build agents and the broader enterprise AI capability.
• Ensure that IT service management leverages AI to deliver measurable ROI while establishing institutional AI muscle.
• Accelerate the SDLC with AI to expand enterprise capacity and increase throughput without increasing headcount.

Together, these moves shift AI from a matter of experimentation to one of infrastructure, from an object of curiosity to an expander of capacity. After watching organizations move from AI pilots to AI programs, I’m convinced that the winners will be the ones that build capacity, not just curiosity. Prospecting may generate headlines, but infrastructure generates durable value. For CIOs, the opportunity is clear: build the railroads.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?