玩偶姐姐

In a legacy industry defined by complex operations and global impact, ConocoPhillips is reimagining energy production through the power of AI. Under the leadership of CDIO Pragati Mathur, the Houston-based company has embraced a strategic, value-driven approach to AI by advancing core priorities like efficiency, safety, and sustainability. From predictive maintenance and demand forecasting, to breakthrough applications like artificial gas lift optimization, ConocoPhillips is delivering tangible business outcomes that align with its core vision.  

What’s your approach to incorporating AI into the overall business strategy at ConocoPhillips? 

As an upstream business, everything we do is in service to producing energy to meet market demand, and our strategy centers on driving efficiencies, reducing costs, and improving safety. We’re also committed to shareholder returns and the company’s ESG goals, and we’re working to reduce our Scope 1 and Scope 2 emissions. Any investment we make with AI, or any other technology, will tie back to those business values. 

We began developing our AI strategy a few years ago in alignment with our overall digital strategy. We’ve been using AI for years but it’s matured to the point where we can reimagine competitive differentiation. We’re seeing outcomes not possible before. 

What are some examples of AI driving outcomes in efficiency, cost, or safety? 

One is predictive maintenance, which allows us to anticipate and fix failures before they occur. We built a complete AI model to predict failures in drill motors, which increases efficiencies in drilling operations. 

Another is with demand forecasting, where we use AI models to leverage historical data and predict market trends. For gas forecasting, we can now predict more accurately both supply and demand for natural gas, including interconnector flows between the UK and the EU. 

But probably the most surprising impact came from a solution we’re particularly excited about: artificial gas lift optimization. Gas lift, which uses injected gas to bring oil to the surface, is not a new concept in the energy industry and we’ve been using artificial lift for years on thousands of wells to produce nearly 500,000 barrels of oil a day. This is a major part of our business.  

In the past, though, our field technicians and engineers would measure and adjust one well at a time, with one engineer responsible for hundreds of wells, so the most they could do was about a dozen per day. However, with the digital telemetry we already had in place for these wells, we’re now using AI as a trusted physics-based calculation to assess every well every day. We’ve increased adjustment rates and lifted more hydrocarbons to the surface, and have sped up the entire flow process. In a commodity business, time is money, so the AI gas lift solution has had significant business impact. 

What were the criteria you used to pursue the gas-lift solution? 

My team has developed a model for use-case analysis, which resulted in the artificial gas-lift solution, and they’re applying it to other areas like reserves forecasting.  

The keys to the model are first to look for the value in our company’s core objectives, and we ask if the proposed solution improves production, efficiency, or safety. A second element of the model is to establish partnerships to build solutions, which is a challenge here, because as an engineering company, we’ve always built everything from scratch. And the third key element is prioritization. The right partner will be as invested in the outcomes as we are, and will really help with prioritization. 

Let’s talk about the team. How have you reorganized your technology organization to function as a use-case generation engine? 

We’re organized into product teams on a global matrixed structure. Product owners are typically not in IT, but are business operations subject matter experts, like reservoir engineers. My team sits on the product team to learn what the business needs, advise on where and how technology can unlock efficiency, and then we execute. Our role is to identify, for example, generative AI opportunities that map to business needs. The product teams are structured to foster partnership and strong collaboration.  

We pull together these product teams from different areas of expertise and ask them to analyze the global landscape, develop the requirements, and make their business case. It’s a Silicon Valley venture capital model, so the idea gets funding or not based on the value of the use case. 

How did you get your board excited about this capability?  

We’re very fortunate to have a tech-savvy board, including members who founded high-tech companies. But even with that, we needed to start with AI 101. We talked about why they should care about it and how it can unlock value for our business. After that, the board asked for deeper one-on-one sessions on AI, and then suggested we go to Silicon Valley to learn more. There, my team and I vetted 100 AI companies, picked 25, and spent two days learning about how their products could help us solve challenges and advance our business.  

The questions we asked every vendor were what’s the art of the possible? What can AI enable? How can it impact speed to market? These are all relatively small companies, but they’re innovating at an accelerated pace.  

We also hosted a technology showcase for the board, where we shared a range of digital solutions. They enjoyed it so much they’ve asked for a repeat with new solutions so they can continue to learn about how we’re applying digital technologies to improve business processes. They now have a deeper understanding of how strategic application of digital technologies can give us an edge and help us stay competitive. 

Storytelling has emerged as one of the most important skills in AI adoption. What was the story you told the board? 

We told them a story about our first use case, for predictive decline curves, which uses historical well-data to forecast supply. We’d been doing predictive decline curves for years, but not with great precision. We sort of predicted and then waited. But with our new AI solution, we have precision in forecasting, and the cost savings have been significant. 

We told the board about the financial impact first, and then when they asked how, we told them this was our AI use case. Rather than spend a lot of time defining AI, we showed them how we used it to solve a real business problem. Our approach was subtle but intentional. 

The 玩偶姐姐 role is constantly evolving. What advice do you have for the person who’ll be in your seat years from now?  

A mentor earlier in my career told me that while I am a technologist, I should read politics, nonfiction, and fiction. Just read a lot and learn to change your mindset. This ability to think differently is one of the qualities I look for in my senior team: learning, curiosity, strategic thinking, and most importantly, a desire to understand how the business works. If you know finance, then spend a day shadowing the operators. If you’re an operator, sit down with finance.  

When I joined ConocoPhillips, I spent time with people in the field, because this business was new to me. I wanted to learn from people across the company and gain an understanding of the full lifecycle. You can’t redesign a business with AI if you don’t know how the business works today. With curiosity, an understanding of your business, and embracing a passion for innovation, you can unlock extraordinary value.