Read enough headlines and you’d think every company is a few months away from running on autonomous agents, but that isn’t the reality for most organizations.

I grew up around this industry in Oklahoma City, and at DataBank we’ve worked with more than 40 oil and gas operators, midstream companies and service firms. The people I talk to aren’t against AI. Most of them are already using it somewhere. What they’re running into is more basic. AI can only work with records you can find and actually trust. For many operators, those records aren’t in shape yet. That gap, between what AI can do and what your records will let it do, is where most projects either pay off or stall.

Everybody wants AI, but the data isn’t keeping up

Oil and gas isn’t dragging its feet on AI. When 451 Research surveyed energy and utilities organizations this year, 27% said agentic AI is already integrated across their organization, compared with 18% across all organization types. Energy organizations have always focused heavily on innovation.

But about a third of the AI projects energy and utilities companies started in the past year have already been killed, compared with about a quarter in other industries. Only 37% are delivering business value. When people explained what went wrong, the top answers were system integration, data quality and data access.

When we talk with people at oil and gas events, they almost never lead with AI. They lead with “I can’t find the record I need,” or “I found it, but I’m not sure it’s the current version.” If your team doesn’t trust the information it has today, putting AI on top of it is like strapping rockets to a garbage can. You’ll go fast, but you won’t end up anywhere you wanted to go. Gartner says roughly the same thing, more politely: it predicts that through 2026, organizations will abandon 60% of AI projects that don’t have AI-ready data behind them.

Every deal makes the records problem bigger

The pile keeps growing, too. Enverus counted $38 billion in U.S. upstream M&A in the first quarter of 2026, the biggest quarter in two years, and expects more consolidation to come. Every one of those deals comes with someone else’s leases, title files, division orders and well histories, usually on paper and filed however the last owner liked it.

Land teams feel it first. One in-house landman told Permian Basin Oil and Gas Magazine in June that he can only use AI about 10% to 15% of the time outside the office, because so much of what he needs was never digitized. No AI tool can fix that on its own.

What it looks like when an operator gets it right

I heard a good example of this in September at the Box Content + AI Summit in Dallas. Nena Calligaro, senior cloud services and infrastructure leader at Merit Energy Company, sat down with our chief revenue and operations officer, Adam Herrmann, to discuss how Merit handles the records that accompany each acquisition. Before 2017, those files arrived in boxes and went onto shelves, and Merit’s engineers and land staff would dig through physical files to find what they needed.

Since Merit started working with DataBank, its land and production records have been digitized, indexed and consolidated into one repository, and files from new deals go through an automated intake process. When an office flood hit, Merit didn’t lose any data. (Here’s the full case study.) Even so, Nena was candid that paper hasn’t gone away: “a lot of the transactions that we see, we’re still seeing a lot of physical files.” That’s the reality for almost everyone in this industry, and it’s why the operators who get value from AI are the ones with a dependable way to bring paper and digital records together as they arrive.

Where to start

The order we see work is pretty simple. Get the records in when the deal closes. Clean them up so there’s one current version of each document, tied to the right well, lease or tract. Then put AI to work on narrow jobs you can check, like lease provision lookups and due diligence questions, and build from there.

This industry will get there its own way

Oil and gas is a show-me industry. People here have sat through plenty of pitches, and they can spot hype from a mile away. Silicon Valley can mistake that skepticism for resistance, but this is the same industry that made horizontal drilling and real-time production monitoring standard practice. It adopts new technology when the technology proves itself in the field. AI will be no different, and it’ll prove itself fastest on records the team already trusts.

So if you’re being asked about your AI plan and you’re not sure your records are ready, start small. Pull the files from your most recent acquisition and take an honest look. They’ll tell you pretty quickly how much work sits between you and AI that actually pays off. And if you want a second set of eyes, our team is happy to help.

Frequently asked questions

Why do AI projects fail in oil and gas?

Most of them stall on data. Energy companies in 451 Research’s 2026 survey named system integration, data quality and data access as their biggest technical obstacles, and they shut down AI projects at a higher rate than other industries.

How should operators handle records from an acquisition?

Bring them in as soon as the deal closes. Digitize and index the files before they go into storage, match them to the assets you bought and merge them into one repository so land, engineering and accounting are all working from the same source.

Sources

• 451 Research, Energy and utilities firms investing aggressively in AI despite mixed returns, March 31, 2026
• Gartner, Lack of AI-Ready Data Puts AI Projects at Risk, Feb. 26, 2025
• Enverus, U.S. upstream M&A hits $38 billion in 1Q26, May 13, 2026
• Permian Basin Oil and Gas Magazine, The Lay of the Land, June 22, 2026
• DataBank, Case Study: Merit Energy