CruxOCM

CruxOCM's whole existence rests on noticing that airplanes have autopilot, refineries have advanced process control, and pulp-and-paper mills have had automation for decades - but the pipeline and midstream sector, moving comparable volumes of comparable value, largely did not. The person who noticed this most acutely was Vicki Knott, who did not read about the problem in a McKinsey deck. She trained as a control-room operator on the Keystone pipeline. She did the forty-step procedures. Then, in 2017, she and Roger Shirt founded a company to automate the job she had personally been doing.

The name is Crux Operations Control Management, which nobody says out loud, so it is CruxOCM. The core technology has an acronym too: RIPA, for Robotic Industrial Process Automation. The idea underneath the acronym is refreshingly literal. If a control-room operator follows a written procedure - start this pump, wait for that pressure, hold this flow rate - then that procedure is, definitionally, a set of rules. And a set of rules is something software can execute. CruxOCM turns the operator's binder into closed-loop automation.

The important design decision - the one that makes the whole thing sellable - is that CruxOCM does not ask an energy company to rip anything out. Its software runs on top of the SCADA and DCS systems the customer already owns and already trusts. There is a genre of enterprise software that fails because it requires the customer to first replace their entire world, and CruxOCM has quite deliberately not built that. It meets the operator at the screen they are already looking at.

The second important decision is that CruxOCM does not fire the operator. This is partly philosophy and partly regulation - safety regulators mandate a human in the loop for exactly this kind of critical infrastructure - but the framing the company uses is the useful one. Knott describes the experience as being "more like adaptive cruise control." The software drives. The human supervises, and can take over instantly. Nobody is proposing a driverless pipeline. They are proposing a pipeline that stops making the human do the tedious, error-prone parts by hand.

What this buys the customer is money, which is why customers buy it. The pitch fits on a bumper sticker: "Same Pipeline. Same Team. More Revenue." You are not building new infrastructure or hiring more people. You are getting more throughput out of the assets you already have, at lower operating cost, with fewer of the mistakes that come from a tired human doing repetitive work at 3am. For a midstream operator whose margins live and die on throughput, this is a very easy conversation to have with the CFO.

The product line reads like a small workforce of specialized robots. pipeBOT handles pipeline operations - starting and stopping pumps, holding flow, watching conditions. gatherBOT does the same for the tangled world of gathering systems. draBOT manages drag-reduction operations. A family of optimization tools - maxOPT, leanOPT, powerOPT - squeezes for throughput, lean running, and energy use respectively. All of it sits inside what CruxOCM calls the Industrial Automation Hub, with the un-glamorous but essential plumbing of deployment management, log analysis, and a real-time status dashboard.

One more thing the company is loud about: it is "physics-first," not black-box AI. This sounds like marketing until you remember what is inside the pipe. When your software is deciding whether to open a valve on a line carrying millions of barrels, "the neural network felt confident" is not an acceptable answer to a regulator, or to a court, or to a control-room operator who has to sign off. CruxOCM leans on physics and explainable logic precisely because explainability, in this industry, is not a feature. It is the license to operate.

The money has followed the logic. In August 2024 CruxOCM closed a $17 million USD Series A - $23.3 million in the Canadian currency it actually banks - led by M12, which is Microsoft's venture fund. The round is more interesting for its guest list than its size. Alongside Microsoft sat ONEOK, a large natural-gas company that is both an investor and a customer, and Raven Indigenous Capital Partners, and the EIC Rose Rock Fund, plus a returning bench of Angular Ventures, Bullpen Capital, Root Ventures, Industry Ventures, Cendana Capital, Pipeline Capital Partners, and Golden Ventures. When a tech VC, an energy major, and an Indigenous capital fund all write into the same round, it tells you the company is not a tourist attraction - it is infrastructure, and infrastructure needs infrastructure investors.

There is a version of this company that goes looking for the flashiest possible framing - full autonomy, no operators, a driverless pipeline splashed across a keynote. CruxOCM has pointedly not built that company, and the restraint is the interesting part. The team describes what it is doing as "Industrial Autonomy," but the arc it draws is a modest, sequential one: first you monitor, then you recommend, then - only when the physics and the regulators and the operators all agree - you execute. Most industrial-AI startups sell the middle step and call it the last one. A dashboard that recommends an action a human still has to perform is a recommendation engine wearing an automation costume. CruxOCM's claim is that it actually does the last step, closed-loop, on live infrastructure, which is a much harder and much more valuable thing to have shipped.

That harder path also explains the customer list, which is short by consumer-software standards and impressive by industrial ones. The company points to work with ONEOK, Enbridge, Delek US, HF Sinclair, and Alliance Pipeline - the sort of names that do not adopt experimental software on a whim, because a bad experiment on a pipeline is not a bug report, it is an incident. When a former CEO of Alliance Pipeline is on record calling the product "proven, scalable, off-the-shelf software to fully automate the execution of procedures," that is not a testimonial you can buy with a discount code. It is the kind of endorsement that only comes after a lot of careful, unglamorous deployment work.

None of this makes CruxOCM a sure thing. Selling software into heavy industry is slow; deployments are careful, because they must be; and the company's own coverage has noted the slightly awkward fact that a Calgary company found many of its early customers south of the border, prompting a Houston presence. But the shape of the bet is clean and old-fashioned in the best way. Find a real, expensive, dangerous manual process. Automate the boring parts. Keep the human. Charge for the throughput. Same pipeline, same team, more revenue - and, not incidentally, a control room that is a little less likely to have a bad night at 3am.