Follow ZDNET: Add us as a preferred source on Google.
ZDNET's key takeaways
- Successful companies refine their strategies to guide AI explorations.
- You can start hundreds of projects, but quantity must become quality.
- Consider key elements, such as architecture, redundancy, and targets.
With many business leaders struggling to create a competitive advantage from AI, your company will need a well-thought-out strategy to beat the rest.
For Art Hu, global CIO at tech giant Lenovo, there are no half measures. “We want AI to penetrate all aspects of our business,” he told ZDNET in a one-to-one conversation at a hotel in London, suggesting his company has created a top-down and bottom-up commitment, where employees are encouraged to explore AI in a tightly governed and secure manner.
Also: 5 ways to prevent your AI strategy from going bust
Key use cases include conversation summarization to assist support specialists, applying agentic AI to enterprise-grade software engineering, and using gen AI to create effective marketing collateral.
Hu and his team have built a careful strategy to shape their AI explorations — and here are five ways you can do the same.
1. Adopt a portfolio approach
Hu said Lenovo takes a portfolio-based approach to emerging technology across the broad AI product lifecycle.
“That approach stretches all the way from ‘I've heard of AI, and I'm thinking about it,' through to ‘I'm playing with it in a sandbox,' to ‘I've deployed it to my department,' to ‘Hey, tens of thousands of people at the company use this tool,” he said.
This portfolio approach means the company has more than 1,000 registered projects across all business areas.
Also: This company's AI success was built on 5 essential steps – see how they work for you
Hu has partnered with the firm's chief security officer and chief AI officer to establish a policy for AI exploration that has shifted from letting thousands of flowers bloom to providing careful guidance.
“We want people to bloom and explore, but we need to control the risk, because there is quite a long tail of unexpected outcomes, especially if you're not careful,” he said.
“You actually want to be in a position where people are very excited, where they're pushing you on approving more projects.”
2. Hone your operating model
Hu said the key implication of encouraging people to explore AI is that IT must be run differently.
“In the past, people used to come to IT, and we'd run a highly centralized architecture,” he said. “The way it translates for us in an age of AI is that we've had to rethink how we work with the business.”
Hu: “We want AI to penetrate all aspects of our business.”
Lenovo
Rather than his team taking non-IT professionals' requirements and turning them into systems, whether through off-the-shelf or self-developed applications, Hu's technology organization must now carefully manage a new area of demand: generative and agentic AI.
Also: 3 ways AI agents will make your job unrecognizable in the next few years
“Instead of a small subset of employees working in IT transformation and building systems, the barrier is lower, and there's potential for the entire company to contribute to digitalization and intelligent transformation,” he said.
“If the whole company wants to get involved in AI and transformation, how do we manage that? It's a very different operating model for the guardrails and guidelines.”
3. Embrace spare capacity
Hu said it's also critical to consider the concept of redundancy in an age of AI. Previously, and probably for most of the past two decades, business leaders prioritized globalization and centralization.
Now, volatility and uncertainty characterize modern economic and technological practices, and that new normal means CIOs must adopt a fresh approach.
“We are architecting and pivoting our systems so that they are going to be much more regional, and that's because of increased focus on data sovereignty and data privacy by the important regulatory regimes from China to the EU to the Americas,” he said.
This new approach to enterprise architecture starts as a business strategy, with in-built shock absorbers that provide redundancy across regional variabilities.
Also: Dreading AI job cuts? 5 ways to future-proof your career – before it's too late
In short, spare capacity that might once seemed wasteful is now a sensible strategy for companies operating in an AI-enabled economy.
“With redundancy, if something happens in one part, your whole system doesn't get paralyzed or become unable to operate,” he said.
“Normally, in the hyper-efficient world, redundancy is bad. But I think through COVID and current volatility, the natural response is, ‘Well, maybe buffers are good,' because you don't know what's going to happen necessarily in the future.”
4. Create a scoreboard
Hu said another key element of Lenovo's AI strategy is that each executive committee member has AI goals.
“So, we're all committed to adopting AI,” he said. “That has an interesting competition effect, because we put up a scoreboard: ‘How is everyone's area doing?' And you don't want to be slower than the person next to you on the executive staff.”
Also: I test AI for a living, and these 3 free tools are the ones I used most in 2025
Hu said the competition is healthy because the resulting dynamic is that professionals in functions, such as marketing, sales, or HR, are all thinking about how to use AI productively.
“We think about our business in a very structured, systematic, and detailed way. We have quantitative targets mapped very specifically to level one, level two, and level three achievements,” he said.
“From a lifecycle and value-chain perspective, we're trying to provide as much transparency as we can. And it's fun if you play the time lapse of lighting up this map of the company and seeing how it's going.”
5. Whitelist great tools
Of course, pace is important when it comes to developing AI initiatives, but so is quality. Hu said that Lenovo tries not to over-prioritize quality initially.
“I'm not going to worry about small amounts of funding at the early stage because we want to build a funnel,” he said.
“Early on, the most important thing is, ‘Did we learn something?' Quality matters a lot more when you want to productionize AI for hundreds or thousands of people. For five or 10 people, if it works and you learn something, that's good.”
Also: Gemini vs. Copilot: I compared the AI tools on 7 everyday tasks, and there's a clear winner
Hu said moving to a focus on quality requires clear direction on requirements and tools. One crucial element for success has been establishing governance to cut through some of the massive uncertainty and noise around AI tools.
“What that approach translates into for us is that we've ended up whitelisting a set of tools to say, ‘We think these will meet 85% to 90% of your requirements if you're starting to explore AI,'” he said.
“But there's no guarantee these tools do everything, so we also have an ongoing process to introduce tools. People can say, ‘Hey, I have a use case, and none of the tools that you're providing are well-suited. Can we get this tool on the whitelist?' And then we'll do the appropriate reviews and assessments to be able to introduce that tool.”
