Last month I spent three days walking the floor at the Global Energy Show in Calgary, talking to the companies bringing AI into oil and gas. Some are names you'd recognize. Others are startups nobody outside their customer list has heard of, quietly running systems inside major operators

Here's what surprised me, a year ago people were still debating whether AI belonged in this industry. The argument has now moved on to who's doing it right, and who's just talking.

Over the next few issues I'll introduce you to the ten most interesting companies I met on that show floor, along with the AI and energy news worth your time. This first issue is about autonomy, meaning AI that takes action instead of writing reports for people to ignore.

Welcome to The Mudlog.

Ahmed

What's moving in AI and energy right now

Chevron is becoming an AI infrastructure company. Chevron and Microsoft signed a 20-year gas deal to power AI data centres, with Chevron developing the Project Kilby power plant alongside Joulent. Final investment decision expected by end of 2026, first power in 2028. An oil major just locked in contracted demand from the AI boom instead of riding the commodity cycle, and if it works, others will copy it.

Alberta has its own version of that story. Pembina Pipeline and its partners just approved the Greenlight Electricity Centre, a 932-megawatt gas plant in Sturgeon County built to power a data centre. Startup is targeted for 2030, and they already hold permits to double capacity. Alberta wants to be the home of Canada's data centre buildout, and the province's gas producers are lining up behind it. Enbridge alone says it's advancing more than 50 potential data centre opportunities. Same pattern on both sides of the border: gas producers turning into AI power companies.

Most operators still haven't pulled the trigger. Bernstein research pegs only 13% of oil and gas companies as having deployed agentic AI, though 49% plan to this year, and Rystad figures digital initiatives could save the sector more than $320 billion through 2030. A lot of money sitting in the gap between the talkers and the doers. Which brings me to the companies in this issue.

AIQ is a joint venture between ADNOC and Presight, based in Abu Dhabi, and they've spent six years building AI for one industry only. Their technology already runs across more than 500 ADNOC wells, covering over 20% of its reservoirs and 100% of its drilling operations.

The numbers got a lot bigger this year. AIQ signed a $340 million contract with ADNOC to roll out their ENERGYai platform across more than 28 producing fields over three years. In a test environment using just 15% of ADNOC's data, a seismic agent interpreted data ten times faster with a 70% gain in precision. Pilots don't produce numbers like that. Production systems do.

The thing that separates AIQ from most vendors on that show floor is how deep the autonomy goes. Their closed-loop well control system adjusts gas injection and choke settings in real time with no operator in the loop. You set the intent, the system executes. They claim to be first in the world to run a system like this in a live field, and given what's deployed at ADNOC, the claim has weight behind it.

Genesis, their newest product, pushes further. It's built to work as a full energy operating system, orchestrating entire oil and gas workflows from one platform using what AIQ calls 100% deterministic AI. The technology stretches across the whole value chain, from subsurface interpretation and drilling through production and asset integrity, and runs in OT field environments as well as cloud-based IT. On the controls side they work closely with Halliburton and Yokogawa.

Having proven the model in Abu Dhabi, AIQ is now expanding into India, Vietnam, Malaysia, and Kazakhstan. They came to Calgary to bring the platform to Canada. Canadian operators should be paying attention.

DataRobot built an agentic AI platform for the mess most large operators actually live with: data scattered across multiple sources, multiple ERP systems, and a multicloud setup accumulated through years of acquisitions and infrastructure layered on infrastructure.

Edge deployment is what makes them worth knowing about in oil and gas. Picture a drilling rig that depends on an agent watching 70-plus channels of real-time drilling data, and the satellite uplink dies. A locally deployed edge agent keeps running through the outage and syncs back up when the connection returns. Same logic for refineries, remote processing facilities, or any operation that can't stop working because the internet did. Running at the edge also cuts how much data needs transmitting in the first place, which matters when bandwidth in the field is thin.

Weeks before the show, DataRobot announced a collaboration with Chevron to apply agentic AI at the edge for autonomous inspection operations at Chevron facilities. That tells you where this is going. The next generation of field AI won't need the cloud to keep working.

DataRobot is newer to the Canadian market and came to the Global Energy Show to build a presence here.

Data2 is a US-based, veteran-owned AI company making one very specific claim: they hold a patent on hallucination-resistant, explainable AI. In an industry where a wrong answer from a machine can end in an operational or safety incident, that claim earns a closer look.

Their platform, reView, pulls structured, unstructured, and time-series data into a single explainable system. Every answer traces back to verifiable evidence through graph-connected knowledge networks, with visual lineage maps showing exactly which data produced each response. Traditional retrieval-augmented generation injects relevant data into a prompt and hopes for the best. Data2's approach makes every answer verifiable before it lands in front of a decision maker.

Both founders of their energy practice carry 20 years of industry experience, and they came to the Global Energy Show to expand into Canada through a partnership with Bow River Solutions, a Calgary data consulting firm that specializes in governance and analytics for regulated industries. Deep energy domain knowledge paired with a patent on hallucination resistance is a strong pitch for any operator wondering where an AI recommendation actually came from.

Issue 02 covers three companies working on the most stubborn problem in oil and gas: data that's scattered, siloed, and hard to act on. Denodo, Data Squared, and Ezops each answer the same question differently. In your inbox in two weeks.

If someone forwarded you this issue, subscribe at themudlog.com to get The Mudlog in your inbox.