An open letter to oil and gas operators · Last updated July 29, 2026
Should we build it in house?
Eleven of the twelve largest US oil and gas operators report no research and development expense line at all (SEC FY2025 10-K filings). That is not a criticism. It is the strongest argument we can make for why you should not be building your own software. This letter makes it in full: the evidence, the real cost of the build path, and the standard to hold every vendor to, starting with us.
The letter
Dear oil and gas operators,
You are not a software company, and you do not want to enter that business.
That sentence is not a comment on your ability. It is a comment on your business, and it is closer to a compliment than anything else in this letter. The market does not ask you for software. It asks you for barrels, cost discipline, and returns, and the best of you deliver those at a level no software company could survive being measured against.
We believe AI belongs in your shop. Use it to reduce friction, automate analysis, eliminate duplicate data entry, and make your people more productive. Use it hard.
Do not use it to try to become a software developer.
At WorkSync we value our partnership with operators and with the technical teams inside them. We know you are smart enough to build apps. Some of you already have, and some of those builds are genuinely good. We also know it is not as valuable as your core focus. The question in the title of this letter deserves a better answer than a vendor's reflex, so the evidence is attached below the signature. The short version: building the app was never the hard part. The hard part is everything after the demo, and it never ends.
We know because we live it. We have spent the last year perfecting our pump by priority platform, and we know it can help the right operators pull value out of their existing operations fast. We ship new versions at least twice a week. We spend tireless hours and a lot of compute hunting down and killing bugs while shipping better and better features. We use AI to make our own people more productive, and the difference is real. That is what the software business demands, every week, forever. It is a good business. It is our business. It does not need to be yours.
But trust us when we say this. Your strategy should be to stay focused on your core competencies and to use the thing you are genuinely world class at: competitive sourcing. Go get the best price and the best terms for your technology services the same way you do for everything else. You already source drilling, completions, and fabrication from specialists, hold them to spec, and keep the returns. Software should not be the exception.
And that stack in the corner costing you millions. Hold it accountable.
ChatGPT has been public since November 2022. That is nearly four years. If nothing in your operation has measurably changed in that time, something is wrong, and it is not the technology.
Now hold us to exactly the same standard.
If you run more than 500 wells, let us put your own production data through our platform. Two weeks, no theater. You will see precisely what your field is leaving on the table, priced in dollars, well by well. If that number is not worth your time, we will be the ones to tell you.
We are rooting for you. We would just rather your best technical people spend next week on your rock, your wells and your field.
The WorkSync team
What the filings say
The paperwork agrees with the letter.
You do not have to take a vendor's word for any of this. The primary sources are unusually blunt.
No R&D expense line
Eleven of the twelve largest US oil and gas operators report no research and development expense line at all. The exception is ConocoPhillips, at $78MM: about 0.127% of revenue. (SEC FY2025 10-K filings)
Hits for revenue from software
A full-text search for revenue from software across all twelve operators’ FY2025 10-Ks returns nothing at all. (SEC full-text search, FY2025 10-K filings)
Where AI lives in the filings
Every mention of artificial intelligence in the FY2025 10-Ks of EOG, Devon, Diamondback, ConocoPhillips, and Occidental sits in a risk factor, not in a strategy or revenue section. (SEC FY2025 10-K filings)
The merger proxy
A 2025 E&P merger proxy ran 1.46 million characters: two investment banks, four peer sets, every standard valuation method. Software scored once, in a boilerplate IP definition. Artificial intelligence scored zero. (SEC merger proxy filing, 2025)
One honest caveat, because it is the standard we are asking you to hold vendors to: the absence of a reported R&D concept in a filing is not absolute proof of zero spending, and one of the twelve claimed $58MM of R&D tax credits in FY2025 (SEC FY2025 10-K filings). The pattern still stands: as a group, the largest operators in the country neither report software development as a material expense nor earn a dollar of software revenue.
Even the most software-flattering metric makes the same point.
The twelve largest US operators run a median of $7.31M of revenue per employee (SEC FY2025 10-K filings, revenue and headcount from the same filing). The median private SaaS company runs about $130K of ARR per employee (SaaS Capital, July 2025). And we will concede the obvious before anyone raises it: the operator number is high because capital and contractors do the work. That concession is not a weakness in the argument. It is the argument. Revenue per employee is the yardstick software companies live and die by, and it does not describe your business at all, which is exactly why entering theirs is a trade down.
The field-grade cost breakdown behind this letter, layer by layer, lives in Build vs Buy AI for Oil and Gas: Where the Cost Actually Lives.
The other side of the ledger
The real cost of becoming a software company.
We know you are smart enough to build apps. That was never the question. The question is what a build obligates you to, because production software is not a project. It is a posture, held forever. Here is what it requires, from a company whose entire job this is.
A shipping cadence you must sustain
We ship new versions at least twice a week, because the models underneath this software move every few months and the field notices when you fall behind. A build that ships quarterly is not a slower version of the same thing. It is a decaying thing.
Bug hunting, funded forever
We spend tireless hours and a lot of compute hunting down and killing bugs before the field finds them. That work never appears in a build proposal, and it never ends.
A security posture, audited
Field software touches production data and sits next to your SCADA estate. That obligates a real security program: reviews, penetration testing, independent audit. We hold ourselves to SOC 2 (Type II in progress, report targeted Oct 30, 2026). A build inherits the same obligation, unamortized.
A team you must hire and retain
Engineers, product, QA, and on-call support, recruited and kept against technology-market compensation. Not a project team that disbands at go-live. A permanent line on the org chart, competing for budget with the people who find and produce barrels.
Your best people, off your rock
Execution is where operator talent pays. BCG’s 2025 Permian Performance Benchmark (19 operators, roughly 60% of basin production) put the top-to-bottom-quartile LOE gap at $8.6/boe in the Delaware and $5.2/boe in the Midland. That gap is closed on wells, routes, and failures, not on sprint plans.
And one thing we want to say carefully, because it is the opposite of an insult. Operators are doing real AI work. Devon's CEO told the market that “the AI revolution is real” on the company's Q1 2026 earnings call (May 6, 2026). The market's response has been just as consistent: it assigns that work zero enterprise value, because operators are valued on barrels, cost of supply, and capital discipline, not on the software they write.
Which sets the only test that matters. An internal build cannot be justified by strategy points or a better multiple, because the market will not pay one. It has to be justified entirely by the operating metric it moves, measured against what a productized platform would move for a fraction of the standing cost. We built our pricing around passing that same test.
The case for partnering
What partnering gets you.
Here we are openly talking our own book, so hold every line to the measured standard: the figures live, labeled, on the numbers page.
A platform proven at scale
The best Pump by Priority platform, measured across 5,000+ wells in live deployments: 15% free cash flow uplift on the same crew and 35% fewer miles driven, both measured. Every figure we publish carries its label on the numbers page.
Shipping you never have to staff
At least two releases a week, the bug hunting, the model porting, and the security program, all carried by us and amortized across every operator on the platform instead of funded alone by yours.
A 4-week stand-up by a team that has done it before
Read-only integration on the SCADA, historian, and accounting systems you already own, with ranked plans in the trucks in weeks. Fit-for-purpose. Plug-and-play. No multi-year integration program, no curriculum.
The Impact Guarantee
We charge when your number moves. The pilot is signed against one operating metric and your own baseline, the same standard we argue every internal build should face.
Competitive sourcing works on us too
Hold us to the same standard you hold every vendor. Price us, benchmark us, put us in a bake-off, and make us win the terms. We built the business assuming you would.
Your technical team stays in the loop
Agents assist in a controlled way; your people make the judgment calls. Your engineers keep the domain knowledge, own the data, and spend their time on the asset instead of the plumbing.
The sibling argument, why adapted enterprise software fails the field, lives on the fit-for-purpose page; what the platform does all day is the complete guide to pump by priority; the unnamed pattern comparison is on Why WorkSync.
The standing offer
Two weeks. Your data. A number.
If you run more than 500 wells, the offer in the letter stands exactly as written: put your own production data through the platform for two weeks, no theater, and see what your field is leaving on the table, priced in dollars, well by well. If that number is not worth your time, we will be the ones to tell you. Want the shape of the math first? The route economics page carries the calculator and its stated assumptions.
Build vs buy, common questions
Should we build our own pump by exception software?
The goal is right and the build path is the expensive way to reach it. The hard part was never the app: it is the always-current data layer underneath, the economic scoring engine, and the permanent upkeep that keeps the field trusting it. WorkSync productized that whole loop as pump by priority, proven across 5,000+ wells in live deployments. Hold a productized platform to your metric before funding a multi-year internal program to reach the same place.
What does it cost to build oilfield software in house?
The first working version is the smallest line item, so it is the number that gets quoted. The real cost is everything after the demo: a standing engineering, QA, and support team, re-porting the system every time the underlying models shift, a security and audit program, and your best technical people working on plumbing instead of the asset. Price the second and third year, not the demo.
Can our internal technical team extend WorkSync?
Yes, and the platform is built assuming they stay in the loop. Integration is read-only on the systems of record you own, your team configures the economics, thresholds, and routing constraints that encode how you run your assets, and agents assist in a controlled way while your people make the judgment calls. Partnering removes the plumbing and the upkeep, not your engineers’ ownership.
How should operators decide build vs buy for AI in oil and gas?
Three tests hold up. Own your edge, not your plumbing: build only what encodes something genuinely proprietary about how you run assets. Price the standing cost, not the first version. And demand the same accountability from both paths: because the market assigns operator-built software zero enterprise value, an internal build must justify itself entirely on the operating metric it moves.
What if we already started building?
Nothing is wasted. Narrow internal tools that automate a process only you own are often good builds, and the integration work your team has done maps directly onto a platform deployment. The honest move is a side-by-side: run your own production data through a proven platform for two weeks, put the result next to your build’s trajectory and standing cost, and let the operating metric decide. Sunk cost should not make the call.
What do SEC filings say about operators building software?
They show no software business anywhere: eleven of the twelve largest US operators report no R&D expense line (ConocoPhillips is the exception at $78MM, about 0.127% of revenue), a full-text search finds zero hits for revenue from software across all twelve FY2025 10-Ks, and the AI mentions of the largest filers sit in risk factors. Honest caveat: an unreported R&D concept is not absolute proof of zero spending, and one of the twelve claimed $58MM of R&D tax credits in FY2025. The pattern stands, and it is discipline, not a gap.
Are your best people working on your most valuable work today?
Hold every vendor to a number. Us first.
One operating metric, your own baseline, and we charge when your number moves.
Continue the cluster: Fit-for-Purpose · Pump by Priority · Why WorkSync