Building Predictive Safety Systems in Energy Operations - with Patricio Rivera of Oxy

Summary

---

Podcast Summary: The AI in Business Podcast

Episode: Building Predictive Safety Systems in Energy Operations
Guest: Patricio Rivera (Former VP of HSE International, Oxy)
Host: Matthew DeMello
Date: May 12, 2026

---

Episode Overview

In this episode, host Matthew DeMello interviews Patricio Rivera, former Vice President of HSE International at Occidental Petroleum (Oxy), to explore how AI-driven predictive safety systems are transforming safety management in the energy sector. The discussion centers on moving beyond reactive approaches—like incident investigation—toward systems that anticipate risk. Rivera highlights the value of learning from successes ("green days") as much as failures, the irreplaceable role of human judgment, and the necessity of aligning safety with business strategy for enduring ROI.

---

Key Discussion Points & Insights

Moving from Reactive to Predictive Safety

• Traditional Focus: Energy organizations have long been adept at investigating incidents (root cause, tap root analysis) to prevent future recurrences.
• The Shift: The next evolution is using technology and AI to predict and anticipate risks rather than simply reacting to incidents.
• Defining Success: Safety isn’t just about avoiding accidents ("safety first") but ensuring business continuity through "safe production"—integrating safety with operational goals.
• Quote, Patricio Rivera (02:53):

  "I personally have matured from safety first to safety always to safe production. Because we are part of a business and we want to be a component of that bottom line performance."

The Importance of Learning from Good Days ("Green Days")

• Observation: Traditionally, the industry has focused more on learning from what goes wrong ("red days") and too little from when things go right.
• Green Day Philosophy:
  • Memorable Insight: "Learning from a good day, learning from a clean day." (07:21)
  • Investigating successful days helps organizations identify which controls, behaviors, and systems are working, potentially extending periods without incidents.
• Pivotal Quote (09:13):

  "I don't think it's less to learn in a green day. There's a lot to learn in the green day. The problem is that we're not learning from that because...the tendency in the industry is to capture the things that are wrong."

"Window of Hazard" & Data-Driven Intervention

• System Development: Rivera describes building AI tools using existing observational data (hundreds of thousands of frontline safety observations) to flag periods of increased risk—termed as entering a "window of hazard."
• Analogy:
  • Similar to wildfire warning signs at national parks, these predictive systems alert users to periods of heightened exposure based on data patterns, prompting proactive intervention.
• User-Focused Design: The tools must translate technical insights into usable, simple information for both frontline workers and executives.
• Quote (10:15):

  "I am a transparent user. Whatever happens in the cloud or whatever happens in numbers, ones and zeros is up to you. I need a product that will help me to secure my business continuity. That has to be practical for the users."

The Enduring Role of Human Judgment

• Humans-in-the-Loop: Despite technology’s capacity to highlight patterns, it cannot replace experience or judgment—especially crucial in high-risk environments.

• Building Capacity: Organizations must equip people at all levels to respond effectively, foster open communication (no blame), and understand contextual human factors that may affect risk.

• Quote (15:30):

  "Technology will never replace experience and judgment from people. That's very important for us at the management level to pass that message to others and also to the people actually in the front line..."

• Human Factors: Acknowledgement that personal issues (e.g., family stress, financial worries) can impact concentration and safety, necessitating compassion and vigilance in safety management.

• Memorable Story (16:40): Rivera recounts photographing an offshore rig to remind friends that energy production is a risky, complex endeavor undertaken continually by real people.

Leadership, ROI, and Embedding Safety in Business Strategy

• Beyond Compliance: An HSE management system alone doesn’t drive lasting results—leadership alignment and integration into business fundamentals are pivotal.

• Communicating ROI: Safety must be positioned not only as a compliance or ethical imperative but as a source of business continuity, financial stability, and even corporate reputation.

• Advice to HSE Professionals:

  • Understand and articulate the financial and operational value of safety to stakeholders.
  • Frame safety initiatives as enablers for long-term performance—not as attempts to create a "world without mistakes," but as systems to manage and learn from them.

• Quote (18:42):

  "The hscms alone isn't enough for lasting safety performance. It will not work by itself. It has to be linked to other fundamentals, for example leadership. You can have the best written management system but if leadership support is not there, it's not going to work."

• AI Beachhead: HSE serves as a strong entry point ("beachhead") for deploying AI pilots within enterprises, given its intersection with multiple business functions and executive priorities.

---

Notable Quotes & Memorable Moments

• On Learning from Success:

  "There's a lot to learn in the green day. The problem is that we're not learning from that..." —Patricio Rivera (09:13)

• On Human Judgment:

  "Technology will never replace experience and judgment from people." —Patricio Rivera (15:30)

• On ROI and Integration:

  "...the hscms alone isn't enough for lasting safety performance. It has to be linked to other fundamentals, for example leadership." —Patricio Rivera (18:42)

---

Timestamps for Key Segments

• [02:53] – Patricio outlines the evolution from safety first to "safe production" fundamentals.
• [07:21] – Introduction of the "learning from a good (green) day" concept.
• [10:15] – Explanation of the "window of hazard" predictive system and user-centric AI tool design.
• [15:30] – Role of human judgment and capacity-building in predictive safety systems.
• [18:42] – Embedding HSE into business strategy and measuring ROI.

---

Conclusion: Core Takeaways

1. Predictive Safety Requires Learning from Success, Not Just Failure:

   • Harness existing operational data to spot trends from both good and bad days.

2. Human Judgment is Irreplaceable:

   • AI can flag risks, but human experience and context remain paramount in making safe decisions.

3. Embed Safety into Business Strategy:

   • Leadership alignment and business-wide integration are vital for lasting ROI and operational stability.

---

For more on practical AI adoption and strategy in global industries, stay tuned to The AI in Business Podcast.

Transcript

[A]

Welcome everyone to the Emerge AI in Business podcast. Today's guest is Patricia Rivera, former Vice President of HSE International at Oxy. Occidental Petroleum is a US based energy company focused on hydrocarbon exploration across the United States and the Middle east, along with petrochemical manufacturing in north and South America. Patricio joins Emerger's Matthew DeMello on today's episode to examine how energy organisations can move from reactive safety practices to systems that anticipate risk. He outlines why learning from good days is as important as investigating incidents. Patricio also emphasizes the enduring role of human judgment, the need for leadership alignment and the importance of integrating safety systems into the broader business strategy. Just a quick note for our audience that the views expressed by Patricia Rivero on today's program do not reflect that of Oxy or its leadership. For our solutions partners, position your brand alongside the Fortune 500 leaders defining the enterprise AI roadmap for the opportunity to showcase your solution to the executives currently funding and scaling global initiatives. Partner with Emerge. Secure your [e-mail] partner that's go.emerj.com partner now the conversation with Patricia.

[B]

Patricio. Thank you so much for being here. It's a pleasure having you.

[C]

Thank you for inviting me to this conversation.

[B]

Absolutely. Especially the energy sector. I don't think we talk to enough on the show just when it comes to regulated industries and where we're seeing the AI impact, but just kind of giving a summary to a few guests that we've had on the show, especially in the HSE area. Despite decades of progress in safety protocols, many energy organizations remain reactive when it comes to health, safety and environmental performance. Incident investigations and audits may explain what went wrong, but rarely help anticipate what might go wrong next. And as those expectations grow for continuous improvement and quantifiable ROI and safety, HSE leadership face the challenge of shifting from compliance to oversight while also aligning safety culture with operational and financial goals. That's what we've heard, at least in a lot of summaries from other guests on the show so far. But just from your advantage point at oxy, what are you seeing as the biggest opportunities to improve health, safety and environmental outcomes in energy today? And how have those opportunities evolved in your experience?

[C]

It's a very interesting question that we ask ourselves very often and you know, summarizing the whole approach is what is different today versus what was in the past? What other components systems technologies can we introduce to our to our management systems or our safety systems in order to make the organization safe. And most of the time we are very good into incident investigations. For example, we do tap root analysis, root cause analysis, depending on the type of company. The companies that may have different approaches, but it's fundamentally the same what happened in the past to avoid the recurrence of this event in the future. So we look into that and see can we do something different. Can we, for example, predict the next event? Can we use technology to help us predicting it? And that is basically the approach that we try to give in order to secure what I believe is one of the most simplistic definitions of safety in oil and gas business continuity. So we have to look into that. We have mature also from the perspective that we are part of a business. Meaning a lot of people, even in my own organization talks about safety first. I personally have matured from safety first to safety always to safe production. Because we are part of a business and we want to be a component of that bottom line performance. If you say safety first, stay at home and then you will be safe. You will never have an accident, at least not in an industry related accident, but you will not be able to produce that product that we sell that is allowing us to support the community and also enjoy the kind of life that we like. So technology forgetting the human factors, because there's also some issues that we're facing as we progress with these ideas. It's the tendency that we tend to say, okay, the technology will be replacing some of the jobs, et cetera. But we also have to consider that portion which is important in the equation of being safe.

[B]

Absolutely. That safe production, as you describe it, sounds a lot like what we've heard. This is an example from our financial services firm friends. They talk about continuous auditing. It sounds like safe production really means continuous safety systems. And that's contrary to safety first in that if, if you're giving 110% all the time, you know, you're going to wear your, you're going to wear yourself out. And it's a little bit more of treating safety as a continuous effort. That's, that's going to need your energy, you know, dispersed over a longer period of time in awareness. Am I understanding that correctly in terms of how you want to approach safety as opposed to this safety first mantra that we've always had?

[C]

Yeah, we, that is, that's the idea. However, we have to be careful in how we articulate our positions because as you said, you know, you can, you never expect to have everything working perfectly. In fact, one of the concepts that we're now using for the development of our programs is the acceptance of certain issues. For example, the fact that people make mistakes or that the systems will influence behavior, or that learning is vital and we have to continue learning as we progress. And also the way we respond to matters, the way we respond to issues, matters a lot. So we have to have those principles that are helping us to mature to the level that the industry, I believe, aiming towards. We have reached certain levels of performance which are great. You know, we learn from experiences we have done, we have introduced more controls, we have done a lot of stuff. But now the challenge, in my view is one to solidify and maintain that performance. And in order to do that, we need to introduce some other tools that I insist technology is helping us to develop without forgetting the humans into our organizations.

[B]

I very much appreciate in your answer, you're not looking for a world without mistakes. You're looking for a world where mistakes don't lead to harm. Do I've got that right? In terms of challenging, kind of the

[C]

status quo view, that is a very interesting statement, Matthew. Yes, we are expecting people to make mistakes. Hence the design of our systems, for example, the design of documents, procedures should consider that that is a critical issue that we have found that we write beautiful documents with technical authors that assume that everything will go as expected, that humans will not make mistakes, the systems will work as per design. And that is not necessarily the case all the time. In fact, one of the issues that I have a lot of energy behind is learning from those days where the system works the way they have been designed for and people behave the way they supposed to. It's what I call learning from a good day, learning from a clean day.

[B]

Right.

[C]

We have that progression. Trying to understand. Go back, let's go back a little bit to the first question. You know, doing incident investigations, we taproot and we learn a lot about that. Maybe wrong, we have to continue doing that. But what happens or what do we learn when we have a good day? And the answer is very little. So we're looking into pushing into understanding what went well in order to get to that point. And if we have this, this set of days, a number of days, a series of days where we have no incidents, I'm calling them a green day. How can we do to expand that tendency and sustain that performance longer?

[B]

Yeah, yeah, Very, very nuanced perspective there. That being said, I know we have a lot of folks come on the show and talk about, you know, yes, there's less to learn from a Green day, as you called it, than a red day. So a bad day. But

[C]

I need to interrupt you there. I don't think it's less to learn in a green day. There's a lot to learn in the green day. That's where I a lot to learn. And the problem is that we're not learning from that because we normally the tendency in the industry is to capture the things that are wrong. You didn't have a supervision in place, you didn't follow the procedure X, you didn't have this lockdown and try device that you were in the line of fire. But when it happens to be right, where is the recording of that good event? And that's something that I can explain if you want the approach that we're giving using artificial technology to point out those things.

[B]

Well, maybe we're diving into there because you spoke about in your first answer, you know, building those in house systems with the data that you've got off the bat. I know in a lot of the preliminary conversations and emails before we turned on the microphones today, you mentioned a window of hazard. Tell us about those systems. Tell us about the window of hazard and what that's trying to accomplish as you're trying to build that in house.

[C]

When all this started with a conversation like this one and challenging the teams, the people in my organization to say what are we learning when everything goes okay? And then we were looking into, the first step is let's define what is a green day. And we had the definition of a green day that has to do, for example, with not having a recordable incident or not having a process safety event and a number of other things. But we concentrate in people first as the first priority. And we were looking into trying to explain, which is another challenging situation with somebody like myself that I have not an IT background. But I told my IT team to say, look, I am a transparent user. Whatever happens in the cloud or whatever happens in numbers, ones and zeros is up to you. I need a product that will help me to secure my business continuity. That has to be practical for the users. And I have. There are different levels of users here. I have users on the front line, important people that are pushing buttons or opening bottles, putting themselves at risk. Very important customer, perhaps the most important one. But I also have to be exposed to a higher, a different level of customers, for example, the board or the executive management, where I have to be able to explain what's happening in simple terms. So simplicity is a value on that approach. Approach. So I gave an Analogy I said, when I go to National Park X, I get into the gate and I see this sign that says today's probability of having a fire is high. And I don't know how they calculate that, but my behavior will be especially careful because I don't want to create an additional hustle for that. So we look into what information do we have in hand. Another important part is that I didn't want to create additional burden over the operation to get more more data. I wanted to use the data that we already captured. My line there has been transforming data and information. So how I transform data information. If I don't transform data information to make decisions, then why we are capturing that data? So what we had is the observations. We have hundreds of thousands of observations. If you talk to anybody that has a safety background observations is self explanatory. I'm observing somebody who does the work. I know that he has to wear a ppe. I know that you can transport the categorize if you wish the safe observations at risk observations. And I was thinking that perhaps when we have a number of safe observations, we appear to be good because again, observations are valid only if there is intervention. When it comes to when you face risk, let's say for example, walking under a load, you have to look into action. Patricia, stay away from that. You are under this law, that's not safe. Say, wait, so you save a life there. When this number of risk observations is coming higher, maybe there is a relationship. Maybe there is something that is a precursor of an incident. Maybe they are telling us carefully because you are breaking all those controls that may lead into an accident. So that's the portion that I'm looking into. And what we have done is we look into the hundreds of thousands of observations and we notice the tendencies and we say it seems that. And I say it seems that because still there's some art in the creation of this algorithm. And also educating the algorithm is another topic that we can discuss. But it's telling me you're getting into this window of hazard. It seems that you are getting into an accident possibility higher. So we want to estimate the probability of having an accident and act upon those things. Because we in theory should know what are the controls that we need to emphasize in order to avoid that to happen. So we have to go to the other side of the equation when we have had accidents to review what is the most frequent barrier or control that has been broken. In order to get to that point and convert to A, you need to work in a B And C controls in order to extend your green day period. That's how.

[B]

Yeah, yeah, absolutely. You had mentioned, you know, building systems around not trying to find a world or build a world without mistakes, but working with mistakes to make sure that. That they're isolated. And you've also, throughout your answers, put an emphasis on the role of humans, just especially for regulated industries, because we hear this all the time, especially financial services, healthcare, even a bit into life sciences, that we're gonna see humans in these roles for a very long time, very, very much on purpose. What do you see as the role that humans are going to play moving forward alongside of these systems, especially as we start fine tuning how they compensate for mistakes rather than trying to make a world without mistakes?

[C]

I think to summarize, the answer is building capacity is what I call building capacity first, building capacity of the company to know how to respond to those issues. Because a lot of the principles is blame solves nothing. You know, Patricia is at fault, but that doesn't solve the issue. We already have an accident. We have to build the capacity for people to speak to us, to all of the management levels, to supervisor levels, to say, here's how we can avoid the situation. That's one thing. The other part is, in other words, technology will never replace experience and judgment from people. That's very important for us at the management level to pass that message to others and also to the people actually in the front line and also understanding that it may be external factors that you cannot control. You know, again, I have to use my name because I don't want to name anybody. Patricio is concerned. It seems that he's not paying attention. But, you know, you perhaps don't know that he's having some issues with the family or maybe he's late on the mortgage and things that may impact your judgment and your concentration to this industry where there is a lot of risk, this is a risk. This is a kind of a funny story. I went offshore and in the way back, I had the chance to take a picture of one of the rigs and you can see the rig and for scale, there was a gentleman down there. The rigs are huge and six, three people still looks very, very small. And I send that to some of my friends. They say for those who think that the energy, the light comes from the wall. So it was just sort of a subliminal message for them to understand that when we do this, there's a number of risks that the industry is taking and, and is taking continuously. But at the same Time continuously improving to, to take the, to protect the life of this worker and others. Protect the environment, of course, and manage risk and continue creating value for our investors.

[B]

Absolutely. In speaking of investors and in speaking of value, it sounds very much, especially from your last answer, like you've had a lot of robust conversations, not least of which about the ROI around human judgment staying in place and then how to best counteract, you know, hey, what's going on in individual employees lives that might lead to more risk. This is a very robust conversation and much more advanced than we hear from, you know, a lot of guests who come on the show and talk about, you know, really managing up to the C suite, especially the expectations around these systems. Wondering if you have any advice for the listeners out there, especially in HSE for the energy sector, about defining and communicating the ROI of HSE initiatives to business stakeholders in a way that's going to make it maybe take them away from the cliff, trying to get a world without mistake.

[C]

And I think perhaps the answer is along the lines of understanding what is the role of an HSE management system and what is the impact of the HSE implementation. One thing that I have told my team members and also when I was progressing my PhD dissertation, I had a lot of information, as you may imagine, for producing that kind of material. The hscms alone isn't enough for lasting safety performance. It will not work by itself. It won't work. It has to be linked to other fundamentals, for example leadership. You can have the best written management system but if leadership support is not there, it's not going to work. Right now the understanding that portion also. I don't want to use the word force, but it's prompting us to as HSE professionals to understand the our business role, the values of financial and non financial concepts. Most of the time we look into, into business continuity to support long term financial stability and corporate reputation. We need to understand that and operations people also need to do that. We have been talking about technology integration and that for me represents a transformative shift in industrial safety. And also the issue about continuous evolution for sustained impact is critical for maintaining safety impact and financial resilience. So the HSE Ms. Has to be a component of the business strategy.

[B]

Absolutely. And that makes, you know, HSE a really, really strong beachhead as we hear from so many guests on the show, a really strong beachhead for AI pilots because you're so connected to so many other functions of the business and you have this front of mind, or at least you should have this front of mind attention for the leadership. I know we're right up on time so we'll have to have you back to talk about that as HSE is a beachhead for AI going forward. But but Patricio, really appreciate you being with here and really diving into these issues that are so important across industries. Thanks so much for joining us again.

[C]

Matthew, thank you for the invite.

[A]

Wrapping up today's episode, I think there are three key takeaways from our conversation with Patricia. First, predictive safety systems emerge when organizations learn from good days as actively as they investigate incidents, using existing operational data to spot early signs of rising risk. Second, human judgment in AI enabled environments remains essential as technology can highlight patterns but cannot replace the the experience, context, awareness and decision making needed in high risk operations. And finally, integrating safety into business performance strengthens long term operational stability requiring leadership alignment. So safety systems function as part of the enterprise's broader strategic and financial priorities. If you have an AI solution, position your brand alongside the Fortune 500 leaders defining the enterprise AI roadmap. For the opportunity to showcase your solution to the executives currently funding and scaling global initiatives, partner with Emerge. Secure your [e-mail] partner that's go.emerj.com partner for further executive level analysis. And to join our network of leaders delivering workflow impact with AI, visit emerge.com on behalf of the team at Emerge. We'll see you on the next episode.