Most VP Ops we talk to want to do AI before they’ve located themselves on the curve. They walk into a quarterly review with a board slide that says “AI initiative, $X budget,” and the question that should have come first never gets asked: where are we today, and what’s the move from here? A four-era frame answers that. It also gives you a permission slip to stop buying things until you know which era you’re in. The eras are how we describe an operation’s relationship to data and decisions, and the gap between the operators pulling away and the ones spending without compounding lives on this curve.
Era 3 is not the safe middle step. It is the most expensive era on the curve. The leaders we work with skipped it.
Era 1, Alarm-driven
The morning starts with a SCADA alarm dump. The pumper drives toward whatever is loudest. The foreman sorts noise from signal in real time, often by phone, often before coffee. There is no economic ranking. A $40,000 deferred-production alarm and a nuisance temperature flutter compete for the same attention because both are red on the same screen. Safety alarms get pulled out of the noise by hand, by people who have done this for fifteen years and know the difference.
Most independents were here in 2015. A handful still are. The signs are specific. You have a war room. You have heroes, the two or three people without whom the operation collapses. You measure missed alarms but you do not measure the dollars on the wells you didn’t visit. You ship a daily plan that is mostly a list of yesterday’s problems. Era 1 persists because alarms are real and ignoring them feels reckless. That feeling is correct. The fix isn’t fewer alarms. The fix is a plan that ranks them.
Era 2, Dashboard-driven
A BI tool aggregates SCADA, production accounting, and a handful of KPIs. Leadership has visibility. The morning report goes to the VP every day. The weekly review is a dashboard. The quarterly is a deck built from the dashboard. Field still decides what to do, often using a fixed weekly route the previous superintendent set up four years ago. This is the most common place we find mid-tier operators in 2026.
Era 2 is a real upgrade from Era 1. You can see what’s happening. You can spot the trend before it becomes a problem. You can answer the board’s questions with numbers. The trap of Era 2 is that visibility and action are separate, and most of the value lives in action. The dashboard tells you 600 wells underperformed by 3% last month. It does not tell you which 60 of those 600 to visit tomorrow morning, in what order, with which crew, against which constraint. Production ops still does that work in the same way it has for twenty years, with experience and instinct and a yellow pad.
The diagnostic for Era 2: ask whether the dashboard changed your foreman’s morning. If the answer is no, you’re in Era 2. If you are paying for several BI tools and the foreman still uses Excel, you are deeply in Era 2.
Era 3, AI-piloted
Specific AI use cases are deployed. Anomaly detection on a tank battery. Predictive maintenance on plunger lift. A document copilot summarizing CMMS tickets. A chat panel grafted onto a SCADA viewer. Each one works in isolation. Each one looks good in a quarterly review. The work loop itself is unchanged.
This is the trap. We have watched operators spend half a million dollars on an anomaly detection pilot that ends up emailing 200 alerts a day to a foreman who already had 200 alerts a day. We have watched operators buy a copilot that summarizes tickets nobody reads. We have watched operators stand up a domain LLM that generates a 12-page weekly report that goes into a folder. The science is real. The product was built well. None of it changed the morning. The pumper’s day still starts with the same alarm dump, just now the alarm has an AI label.
Era 3 is expensive in three ways. First, the direct spend on tooling. Second, the consultants and integrations to stand the tooling up. Third, and worst, the org friction created when an AI program ships without improving the metric anyone cares about. After eighteen months in Era 3 with nothing to show, the board concludes that AI doesn’t work in oil and gas, and the next operator who wants to actually fix the workflow has to fight that conclusion before they get the budget.
The diagnostic for Era 3: did your foreman’s morning change? If a new AI tool shipped and the morning routine is unchanged, you are in Era 3, regardless of how much was spent or how many vendors are involved.
Era 4, Agentic
The work loop runs on agents. Detect, score, route, execute, learn, every shift. The morning isn’t the foreman calling the field; the morning is a ranked plan delivered to every truck cab by 6 AM, scored on cash flow and risk, with safety as a hard constraint, not a weight that can be traded off. The foreman approves, not dispatches. The VP sees plan-versus-actual on Command View instead of waiting for the month-end deck.
Era 4 is where the strategy houses (BCG, McKinsey, Deloitte) keep sizing agentic-AI adoption as a meaningful double-digit profit lever over a five-year horizon, with the upper end of their ranges getting attention because the upper end is what compounds. We see it in the field as 15 percent free-cash-flow uplift on the same headcount, 35 percent fewer site visits, TRIR moving from 1.8 to 0.3, all on the same wells, the same crews, and the same equipment. The variable that changed was the plan.
The diagnostic for Era 4: a foreman who used to spend two hours every morning on the phone now spends fifteen minutes reviewing a ranked plan and approving exceptions. Field crews leave the yard knowing exactly what to work on, in what order, and why it matters. The superintendent calibrates the scoring model on Friday afternoon instead of building the next week’s plan. Engineering stops re-keying GIS data into simulators because the model auto-builds.
The diagonal cut, Era 2 to Era 4 directly
The leaders we work with skipped Era 3. They moved from dashboards straight to agentic by changing the morning workflow, not by buying a copilot. This is counterintuitive because Era 3 looks like the safe middle step, the place where you de-risk by piloting. It isn’t. Era 3 is the most expensive era because every dollar you spend modernizes the tooling without modernizing the outcome.
The diagonal cut is to recognize that the bottleneck isn’t a missing AI tool. The bottleneck is the work loop itself. Once you fix the work loop, AI use cases stop being side projects and start being the engine that runs the loop. Anomaly detection earns its keep because the anomalies feed a ranked plan. Predictive maintenance earns its keep because the predictions are scored against cash-flow impact. The agents have a job because the work has a structure they can plug into.
The cheapest path to Era 4 is the path that doesn’t go through Era 3.
How to assess your operation today
Most VP Ops we ask think they’re in Era 3. Most are in Era 2. The reason is human. If you’ve sponsored an AI project in the last 18 months, you remember signing the PO and you remember the kickoff. You don’t remember whether the foreman’s morning changed because that wasn’t the metric anyone tracked. The honest assessment requires looking at the work, not the tooling.
We built an eight-question assessment that targets exactly that. Each question asks about a specific morning behavior, not a vendor you bought. The questions take five minutes. The on-screen result is free, your era plus the one diagnostic that locked it in. The polished PDF and the personalized 90-day path from where you are to Era 4 gets emailed.
Where is your operation on the four-era curve?
Eight questions. Five minutes. Honest answer. On-screen result is free. PDF + personalized 90-day path emailed when you ask for it.
Hype vs. reality, what AI actually does in the field today
The use cases that are working at scale, the ones that look good in a deck but fail at the wellhead, and how to tell the difference before you sign. Drops next week.