All Posts

The hard problem in drilling automation has moved from the algorithm to the interface. H&P released a standardized RSS-downlink API schema to the D-WIS interoperability workgroup, Mazama Energy reported an 80% cut in drilling days on its Athena geothermal well, and Baker Hughes, SLB, and Nabors each placed integration bets across mature-field AI, offshore CO2 storage, and deep geothermal.

WITS, WITSML, ETP and OSDU each do a different job for drilling data. Here is what each one standardizes, and why the data layer, not the algorithms, is the real foundation beneath drilling autonomy.

The through-line this week: named commitments are replacing pilots. bp awards Halliburton an integrated drilling and evaluation contract with LOGIX automation on the Bumerangue appraisal offshore Brazil; NOV and Well ID report real-time borehole imaging and caliper data through wired drill pipe; and FutureOn, Elementz and DNV move integrity work into structured decision tools.

On most rigs the bit talks to surface at a few to a few dozen bits per second. Wired drill pipe lifts that telemetry ceiling: what it enables and its cost.

ExxonMobil has set a target of putting robots on roughly half its Permian rig floors by 2028, and its first automated rig drilled the third-fastest well in company history. Edition #14 works through what that ramp proves, what it does not, and the day-rate step-down clause to put in your 2027 rig contracts.

Every real-time drilling model, from torque and drag to hydraulics to the rig-state classifiers feeding the digital stack, needs to know exactly what is in the hole. That answer lives in the bottomhole-assembly record, and on most rigs it does not arrive until the company man closes the morning report, hours after the string is already downhole. The models run anyway, on assumptions, and every answer inherits the uncertainty.

Identifying what a rig is doing from its data is really two different problems. Bit rotation cleanly separates washing from reaming, but nothing in the surface data separates reaming from hole enlargement, only the bottomhole-assembly record does. Mislabel either, and every KPI, performance number, and machine-learning model built on that label inherits a silent error.

Four new ways AI and automation entered drilling this month, above the rig, on the rig, and before the spud. WillCo maps each one and what it does, and does not yet, prove.

Agentic AI is already in commercial operation on active drilling programs. WillCo breaks down what autonomous drilling actually means in 2026, where operators are deploying it, and an 8-step roadmap for evaluating and deploying agentic AI on your program.

This Edition at a Glance ADNOC's AD-300 enters live commercial operation as the first AI-enabled walking island rig to execute closed-loop autonomous drilling without continuous driller override on a producing field. Halliburton consolidates LOGIX and DrillTronics into a unified autonomous drilling stack, validated in Guyana with ExxonMobil in the industry's first fully integrated closed-loop automated well placement. SLB reports 2.4 million feet of automated drilling footage

Drilling mud is the blood of the well Think of a drilling well as a living body. The mud pumps are the heart. The wellbore is the circulatory path. The drill string and annulus are the arteries and veins. And drilling mud - the fluid that circulates continuously from surface to bit and back again - is the blood. Like blood, mud does more than move. It carries hydraulic energy to the bit, transports drilled cuttings back to surface, cools and lubricates the bottom-hole assembl

Pure machine learning has structural limitations in drilling that no amount of data can fix. This article explains why - and what physics-ML hybrid models do differently.