Javier Zayas Photography/Getty Images
CIO Rom Kosla’s summary of the importance of emerging technology to Hewlett Packard Enterprise (HPE) likely resonates with any senior executive: “AI is on our mind.”
Research suggests Kosla is far from alone. More than three-quarters (78%) of business leaders report their organization uses AI in at least one business function, according to a recent McKinsey study.
Also: 4 ways your organization can adapt and thrive in the age of AI
Kosla told ZDNET that HPE uses third-party applications with built-in AI capabilities and has spent the past 18 months developing an internal chat solution called ChatHPE, a generative AI hub used for internal processes.
Here are four ways you can use Kosla’s experiences to turn AI into a business advantage.
1. Establish an AI strategy
Like other blue-chip enterprises, HPE uses multiple external AI models and technology platforms. Professionals use Microsoft Copilot to boost productivity, while developers use GitHub Copilot.
“The goal here is that, as we deploy licenses to individuals, they leverage these technologies to do their day-to-day jobs,” Kosla said.
The strategy becomes more interesting around ChatHPE, a bespoke platform that is powered by Azure and OpenAI technology.
Also: This free Google tool turns AI into your research assistant
Kosla said the company has a pipeline of use cases that are assessed and evaluated. For example, a legal team may want to use AI to review contracts, extracting specific insights and generating new templates.
“What we then do is work with them, leveraging ChatHPE, and pointing it to specific source data,” Kosla said. “We’ll also limit access to data, because we don’t want people to use information without care and consideration.”
He also outlined how the use cases might bubble up in other areas of the business.
“In the case of operations, it’s more around chat. You can think of it as something for customer service. For example, there might be a use case where people want to find a part number. They use ChatHPE to access the information, and the technology provides insight to help customers,” Kosla said.
He added that marketing may use ChatHPE to identify what campaigns could be reused, while finance could opt for analytics use cases.
2. Train everyone to use AI
Kosla said his staff often discusses the concept of majors and minors on their journey to becoming a mature IT organization.
“If you think about a university, maybe you’re focused on economics, and that’s really what you want to do,” he said. “However, it doesn’t hurt to learn finance, marketing, or do a minor in computer science. All of those things create a wholesome view of opportunities that you can apply.”
Also: Is your business AI-ready? 5 ways to avoid falling behind
According to Kosla, that’s the approach HPE takes to AI. The company wants all employees to consider and be competent in various majors and minors as skill areas.
“If your major is in supply chain, you should minor in AI. Or if your major is in AI, you should minor in supply chain,” he said. “You’ve got to be as diverse in your knowledge as possible, because everything we’re doing is going to cross the boundaries of other areas of the business.”
Kosla explained how HPE’s concept of majors and minors is bolstered by AI-focused learning and development. “We enable that training with our HR systems. People can come in and learn. The goal here is to make AI training available to everyone,” he said.
But this focus doesn’t stop at training — Kosla emphasizes how important follow-through after the fact is, too. “As they adopt the technology, my rule is always, ‘If you learn it, you should use it.’ It’s not great to have a certificate and be unable to apply the knowledge.”
3. Put guardrails in place
Kosla said it’s already clear that AI will permeate every organization, function, and role, leading to a commensurate impact on information access. Smart business leaders will focus on managing the relationship between agentic technologies and their human counterparts.
“From an IT standpoint, I think of controls and guardrails,” he said. “You need the human factor of understanding questions like, ‘What is the machine doing, and what decisions are they making on your behalf? How are you ensuring that those guardrails are tweaked, and either tightened or widened, depending on the scale of what you’re trying to do?'”
Also: The top 20 AI tools of 2025 – and the No. 1 thing to remember when you use them
Kosla said these guardrails can only be established and refined by IT partnering effectively with the business.
“A lot of the work initially was around questions like, ‘How do we make sure that we manage the prompt engineering correctly, that the data that’s being extracted is limited, the amount of training is managed correctly, and the responses from the engine are accurate?'”
He continued: “There are key phrases we don’t want our engines to respond to because, potentially, the answers could be dangerous or harmful. So, what we’re trying to do is ensure that the prompting is for business usage, but also that the work doesn’t generate too many queries, because we want to avoid large costs.”
4. Consider where you’ll go next
Kosla said HPE’s internal approach to AI continues to mature, and the company will mix in-house and external models according to requirements. The company’s current focus is on using AI to augment staff’s operational activities.
However, Kosla also recognized that AI agents are evolving quickly. His company already uses software from other providers, such as SAP and Salesforce, that are implementing agentic services.
Also: 10 strategies OpenAI uses to create powerful AI agents – that you should use too
One of the key questions HPE is considering is the orchestration of AI across its end-to-end operating model.
“If we orchestrate between two providers and they have two different agents, but an individual is using both and they pivot their chair between one application and another, they might find they don’t talk to each other,” he said. “So, we’ve got to build that orchestration layer, or we have to find the best practice around that issue.”
Kosla said these technical issues will temper some of the hype around agents.
“When you think about an individual, they’re not tied to one application,” he said. “And that’s the challenge we’re facing. There’s a lot of selling when it comes to agents. But you’re not an agent that only does finance — you’re also doing finance, quoting, pricing, and all those systems interact together.”
