5 Ways You Can Stop Testing Ai And Start Scaling It Responsibly In 2026

Andriy Onufriyenko via Moment / Getty Images

Follow ZDNET: Add america arsenic a preferred source connected Google.

ZDNET's cardinal takeaways

• CIOs want to move from AI pilots to enterprise-wide deployment.
• AI must standard safely, responsibly, and pinch clear, measurable impact.
• Digital leaders should prioritize infrastructure, skills, governance, and value.

The adjacent 12 months will beryllium erstwhile the endeavor gets superior astir its AI deployments, pinch tests and explorations being replaced by production-level services that thrust lasting business transformations.

That's nan conclusion of Lenovo's CIO Playbook for 2026, produced pinch tech expert IDC. The playbook is simply a research-led guidebook for integer leaders who want to move from AI experimentation to enterprise-scale deployment.

Also: 8 urgent updates your IT playbook needs to past nan AI era

Lenovo's European CMO Alberto Spinelli explained to ZDNET during a media briefing that nan research, which surveyed 800 executives from Europe and nan Middle East, demonstrates really organizations are moving from scattered take to enterprise-wide strategies supported by stronger foundations successful data, skills, infrastructure, and governance.

"AI is nary longer conscionable a early ambition, and it's now much of a defining unit successful really enterprises operate, compete, and grow," said Spinelli, suggesting that businesses look a captious moment.

"The title is on, but it's not conscionable astir who adopts AI fastest, but who scales it safely, responsibly, and pinch clear, measurable business impact."

Also: 10 ways AI tin inflict unprecedented harm successful 2026

During nan event, Ewa Zborowska, investigation head astatine IDC, summarized nan investigation findings and discussed 5 ways CIOs tin standard their AI initiatives effectively.

1. Put AI astatine nan halfway of business

Zborowska said nan investigation shows that AI is important for organizations seeking faster maturation and amended outcomes.

"AI is becoming not conscionable 1 of nan technologies that customers are trying to usage and thing that they study about, but an important enabler that helps them alteration business, toggle shape really they operate, thrust growth, and build competitory advantage."

That's an attack that Art Hu, world CIO astatine Lenovo, has adopted astatine his organization, arsenic he explained to maine precocious successful a one-to-one question and reply for ZDNET.

Also: Nervous astir nan occupation market? 5 ways to guidelines retired successful nan property of AI

Zborowska encouraged different CIOs to activity pinch their peers crossed nan statement to move their business priorities into clear AI usage cases pinch owners, KPIs, and timelines.

"Yes, you are a CIO, but you are a partner for business. You request to recognize that AI is not conscionable to beryllium considered from a exertion perspective; you request to look astatine it from nan position of what it brings to business," she said.

"Making judge that you look astatine processes and workflows internally to thief adopt AI, and being a beardown partner for your business stakeholders, is going to beryllium of nan utmost importance."

2. Identify impervious of worth

Rather than conscionable utilizing AI to amended tech-based activities, arsenic organizations had predominantly successful earlier years of nan report, Zborowska said much organizations are now utilizing AI to enhance, innovate, and reinvent their businesses.

"The mostly of customers are measurement past this clip of testing and trying things, and overmuch much successful a infinitesimal wherever they think, 'OK, really do we make this caller exertion activity for america successful nan discourse of business outcomes?'"

The investigation suggested that almost 60% of companies are piloting aliases systematically adopting AI.

Also: Weaponized AI consequence is 'high,' warns OpenAI -- here's nan scheme to extremity it

Zborowska said cardinal business priorities for deploying AI see expanding gross and increasing profit, improving customer acquisition and satisfaction, and boosting worker productivity.

"AI implementations are now focused connected things for illustration being much businesslike successful operations and looking for caller ways of generating business," she said.

"This alteration doesn't mean nan things that CIOs talked astir past twelvemonth are nary longer important. It simply shows that different elements are of utmost liking to nan business astatine this point."

3. Scale your infrastructure

However, Zborowska recognized that delivering worth from AI comes pinch challenges. She said nan investigation highlighted communal issues, specified arsenic training and upskilling successful IT teams and business functions.

However, she suggested nan astir important situation is IT infrastructure, stating that AI will not tally efficaciously without businesslike foundations.

"We request to deliberation astir really we present AI into our organizations to guarantee that group are really happy erstwhile moving pinch nan technology," she said.

Also: 4 caller roles will lead nan agentic AI gyration -- here's what they require

Zborowska said conversations pinch CIOs propose that scaling AI is only imaginable if nan business ensures beardown foundations for security, skills, and IT infrastructure are successful place.

The investigation reported that 82% of organizations will leverage on-premises aliases separator deployments for AI workloads and applications arsenic portion of a hybrid environment.

"You must guarantee AI is integrated wrong a wider IT property connected your extremity successful nan astir unafraid measurement and besides put successful spot each nan mechanisms that let you to make judge that you are operating and managing this infrastructure smartly, some from a exertion perspective, but besides from a money perspective."

4. Manage agentic concerns

Zborowska said liking successful agentic AI is rising quickly, pinch nan investigation reporting a 65% summation successful organizations preparing for take arsenic they activity to automate analyzable processes.

The investigation suggested that early attraction areas for agentic AI deployments see information operations, financial workflows, and customer service, domains wherever system tasks make agents effective.

However, organizations still look awesome agentic challenges, specified arsenic ensuring information quality, perfecting workflow redesign, establishing power mechanisms, and managing autonomy.

Also: 96% of IT pros opportunity AI agents are a information risk, but it is deploying them anyway

Zborowska said nan emergence of agentic AI places caller demands connected integer leaders.

"One of nan astir important roles connected nan CIO broadside will beryllium to make judge that you are helping nan business place wherever introducing agents and agentic AI makes sense, and wherever much accepted approaches will still beryllium bully enough," she said.

"Another cardinal information for CIOs is looking astatine agents from nan position of really they will beryllium managed to support control, support security, and forestall agents from sprawling."

5. Govern responsible AI

The investigation reported that conscionable 30% of CIOs person established AI governance policies, rules, guidelines, and practices that rigorously reside security, information protection, privacy, and AI sovereignty.

Worse still, much than half (54%) person not established aliases are successful nan process of processing an organization-wide attack to AI governance.

Also: How these authorities AI information laws alteration nan look of regularisation successful nan US

Zborowska said successful enterprises work together connected clear, shared rules for responsible, transparent AI use.

"Introducing AI into your statement aliases really you activity pinch your customers aliases your business partners needs to beryllium based connected trust," she said.

Zborowska said CIOs execute this assurance by ensuring that AI governance and take return spot manus successful hand.

"That intends ensuring your statement is not conscionable focusing connected 1 element, for illustration information aliases processes, but making judge that each elements are successful place. So, you person rules, policies, and processes, and guarantee that group are besides well-prepped, truthful they're upskilled continually."