AI Isn’t Failing, Your Strategy Is: How to Unlock Real Business Value

Artificial Intelligence (AI) is often misunderstood. Many leaders think of AI as a magic bullet – a quick fix to boost efficiency across their entire business. In reality, AI is more like a toolbox, and the value it generates depends on how it’s used. We had the chance to speak with Patrick Bangert, VP and Chief of AI at Oxy, a global energy company and one of the largest oil producers in the US. With over 20 years of experience in the energy industry, Patrick oversees hundreds of AI projects at Oxy, ensuring they deliver true value to the business. In this article, he shares his insights to help you avoid common pitfalls and maximise the impact of your AI efforts. After all, poorly planned strategies can derail even the most advanced AI initiatives.

Focus on business needs, not just technology. 

Imagine calling a plumber to your house without first having a plumbing issue. As Patrick puts it, “You’d never say, ‘Let’s bring a plumber over and see what they can do.’” 

“Yet, that’s exactly how many companies approach AI. They create an AI department, run workshops, and brainstorm use cases without any real understanding of what’s needed.”

As a result, Patrick estimates that eight or nine out of ten AI projects fail to deliver value – and that’s not because the technology is flawed, but rather due to a lack of alignment with business needs. 

“The failure of AI initiatives is almost never about the software, mathematics, or the data,” Patrick explains. “It’s a failure to achieve value because either you’re doing something that’s just not needed or much more popularly, you’re doing it in an ivory tower without considering the end user.”

In short, Patrick emphasises that you must “Identify the real problem that needs solving and recognise when AI is genuinely the right solution – not because you’ve forced it to fit, but because it truly is the best fit.”

Rank where AI can have the biggest impact. 

With 200 AI projects currently underway at Oxy and another 400 on the horizon, Patrick understands the importance of prioritisation. “We rank projects based on their value – what’s the impact if we don’t do this? How valuable is it if we do? Does it generate extra revenue, reduce costs, or save time?”

But assessing value isn’t a one-size-fits-all approach. For instance, at Oxy, costs are largely tied to assets, whereas at companies like Meta and Microsoft, people-related costs like salaries are a much bigger factor. This means AI projects must be assessed differently. While improving team efficiency may be a priority for some businesses, at Oxy, AI initiatives that enhance asset management – such as reducing operational downtime or predictive maintenance for critical machinery – hold far greater value.

Consider the user at every stage.

Patrick emphasises the critical importance of not just identifying problems but fully experiencing them. “Investigate what the problem is in real life. Experience it firsthand in the field,” he advises. By spending time with the end users and understanding their day-to-day challenges, you can build AI solutions that are not only sophisticated but also practical and intuitive.

To achieve this, you must design AI solutions from day one with the end user in mind. As Patrick explains, “It’s essential to understand what their day looks like and to let them help you shape what the technology can actually do for them.”  For instance, say you’re designing a chatbot for field workers. By going out there and experiencing the problem, you’ll quickly realise that a verbal interface is key, as your colleagues spend the majority of their time driving and cannot interact with a screen. This user-first approach ensures that AI aligns with real-world needs, driving genuine value.

Critically assess potential gains.

When it comes to securing stakeholder buy-in, Patrick stresses the importance of critically assessing AI’s true impact and describes himself as conservative in his approach. 

“Say, for example, you want to use Generative AI to pre-fill a form and save your staff time. You estimate it will save 11 minutes per day, and when you factor in hourly wages, it seems to generate value. But in reality, if staff use the time saved for an extra coffee break, you’ve gained nothing. You’ll only see real value if you can save 10 times that amount.”

Being upfront about the realistic potential of AI initiatives. By not inflating expected gains you’ll build trust, manage expectations, and demonstrate a clear understanding of value. By setting achievable, evidence-based expectations, you strengthen your case and gain the support necessary to drive AI projects forward.

The path forward

If your business is looking to succeed with AI, the lessons are clear. First, define your problems before seeking solutions. Second, prioritise projects based on measurable value. Finally, view AI as a toolbox to empower your team, working closely with end users to ensure it meets their real needs. Doing so will unlock AI’s true potential and drive exceptional results for your business. Approach it blindly, or merely replicate what others are doing, and you’ll set yourself up for failure.

Looking for more insights?

Get exclusive insights from industry leaders, stay up-to-date with the latest news, and explore the cutting-edge tech shaping the sector by subscribing to our newsletter, Commodities Tech Insider. Interested in being featured in a future spotlight? Reach out to [email protected].