AI in Oil and Gas: Do More with Less, Not Less with Fewer

Every investor meeting in oil and gas right now has the same question sitting in the middle of the table: what are you doing with AI?

Philip Richard hears it constantly. He’s the CEO of Porosity, a field workflow automation company, and he spends a significant part of his time talking to investors, prospective customers, and the people building the tools. And what he’s noticed is that the conversation around AI changes month to month, sometimes day to day. But the underlying question from the investment side is almost always the same: can I replace people? Can I replace costs? Can I replace processes?

Philip’s answer is more honest than most. Oil and gas isn’t really about replacing people. It’s ripe for creating efficiency and cutting costs. And those are two very different things.

The real question isn’t whether AI belongs in the oilfield. It’s whether you’re building toward something real, or just chasing a buzzword.

The Cycle Problem That AI Actually Solves

Philip made a point on the podcast that deserves more attention than it usually gets. Oil and gas is a cyclical industry. That’s not new. But what’s changed over the past decade or so is that the swings aren’t as heavy as they used to be. Companies are trying to keep things steadier. Hiring less aggressively in the booms. Cutting less dramatically in the busts.

That sounds like progress, and in many ways it is. But it creates a new kind of pressure. In a low commodity environment, a smaller team might be enough. But in a high commodity environment, when the work ramps and the pace picks up, how do you do more with the same team you already have?

That’s where Philip sees AI making its mark. Not by replacing the people. By giving the people already there the ability to cover more ground, make faster decisions, and handle the kind of workload that used to require twice the headcount.

As he put it: doing more with less within any size oil and gas company. That’s the thesis. Not automation for the sake of automation. Efficiency that lets operators survive the downs and scale in the ups without burning people out.

You Can’t Just Slap AI on It

One of the most refreshing things Philip said in our conversation was blunt: you can’t just slap AI on something and expect it to do something. He and his CTO share that conviction. They’ve watched companies try it. They’ve seen the results. And they refuse to go down that road.

He gave a specific example. A lot of companies look at their IoT sensors and detection devices and think that’s enough to run an AI model. But unless you know what you’re actually trying to achieve, and unless you have human-validated data that defines what normal conditions and abnormal conditions actually look like, you’re just spinning in circles.

Philip’s framework is straightforward. AI has to start with the outcome, not the model. Then it needs clean, aggregated, human-verified data. Not random data sets. Not a sensor feed with no context. Structured information that someone confirmed before the algorithm ever touches it. Then, once that foundation exists, you can start layering in IoT and building toward something predictive.

That sequence matters. And it’s the step most companies skip.

What a Decade of Bad Data Taught the Industry

Philip has been in the industry for about twelve years, and he remembers the wave of data analysts getting hired to study shale data. The idea was sound. Look at how much proppant you’re pumping, how long you’re drilling laterals, what production looks like across different wells, and start finding correlations. Run multivariate analysis. Build a model. Get smarter.

The problem was that it almost never worked. One data set came from one vendor. Another came from somewhere else entirely. There was no cohesion across the inputs. Companies would change four variables between wells instead of one, so there was nothing to isolate. Well pacing wasn’t even being accounted for. The whole exercise was dead before it started.

Philip’s view is that this is actually where AI is making the biggest lift right now. Not in predictive modeling, which he thinks the industry is still a ways from doing well. But in taking multiple, disconnected data sets and smoothing them across each other so you can start building real insights. Making sense out of large, messy data. That’s not glamorous. But it’s where the value actually lives today.

Efficiency Is a People Story, Not a Technology Story

What comes through strongest in Philip’s perspective is that the technology is never the point. The person who benefits from the technology is always the point.

He talked about what’s been stacked on field operators over the past several years. A lease operator used to collect gauges. Now they’re responsible for maintenance, safety, compliance, and reporting on top of their regular work. Environmental regulations, investor pressure, tighter margins. All of it flows downhill to the person standing on the lease.

Philip told a story about a lease operator he knew early in his career who would get out to the field at six in the morning, come home at four in the afternoon, and then spend three more hours at his computer entering reports. Three hours of his evening. Time he should have been spending with his family.

That’s the cost that doesn’t show up on a balance sheet but shows up everywhere else. In burnout. In turnover. In data that gets penciled in by someone who’s exhausted and just trying to be done.

When Philip talks about AI and efficiency, he’s not talking about a dashboard upgrade. He’s talking about giving that operator his evening back. He’s talking about the maintenance platform Porosity released earlier this year where customers can now chat directly with their data through an AI system. Ask it to show the highest cost projects for the past year. Ask it which components failed the most. Minutes instead of hours. Real answers instead of stacking pivot tables.

The results from their LDAR platform tell the same story. Leak repair time dropped from a week and a half to a single day. Errors went down ninety percent. Sixteen thousand man-hours of data entry eliminated over a year. Monthly reports reduced to a single click.

That’s not AI hype. That’s what happens when you build tools that actually respect the people using them.

Philip Richard’s Take on Getting AI Right in Oil and Gas

Philip isn’t pitching a silver bullet. He’s describing a philosophy that’s grounded in how this industry actually works. Here’s what stood out from the conversation.

Start with the outcome, not the model. Too many companies start by asking what AI can do. Philip says the first question should be what are you trying to achieve. Define the outcome, then figure out whether AI is the right tool to get there. If you start with the model, you’ll build something that looks impressive and solves nothing.

Human-verified data comes before everything else. Philip is a big believer in starting with data that a human being has confirmed. Not raw sensor feeds. Not auto-generated logs with no context. Validated, structured data that tells the model what normal looks like and what doesn’t. Without that, predictive AI is just guessing with confidence.

Don’t chase AI to impress investors. Build it to serve operators. Philip was candid about this. AI is the number one topic in the venture capital world right now. But he and his CTO refuse to slap AI onto their product just to check a box. If it doesn’t genuinely help the customer, it doesn’t ship. That’s a discipline more companies need.

The field is where efficiency lives or dies. Porosity’s sales process often starts with the field, not the office. Philip goes to the production superintendent or foreman who’s overwhelmed with data entry and shows them what their day could look like. When the field gets excited about a tool, adoption follows naturally. If you’re trying to push efficiency from the top down without the field’s buy-in, you’re fighting a losing battle.

The companies that don’t adapt won’t survive the next downturn. Philip didn’t sugarcoat this. If operators keep doing things the old way in a low commodity environment, some of them won’t exist. Their companies will get bought. They’ll go bankrupt. The tools are out there. The question is whether leadership has the willingness to use them before the market forces the decision.

Final Thought

The AI conversation in oil and gas has been hijacked by two camps. On one side, there are people selling the future like it’s already here. On the other, there are people pretending none of this applies to them. Philip Richard sits in the middle, and that’s why his perspective matters.

He’s not promising that AI will solve everything. He’s saying that if you start with the right foundation, clean data, clear outcomes, and tools built for the people actually doing the work, then AI becomes what it should be: a way to do more with less without asking your people to do more with nothing.

The oilfield has always run on the backs of people who show up early, stay late, and figure it out. AI doesn’t change that. It just means they shouldn’t have to spend their evenings entering the same data twice.

Join Philip Richard, CEO of Porosity, on Wisdom at the Wellhead as he breaks down why AI in the oilfield has to start with clean data, honest outcomes, and respect for the people doing the work.