WillCo Drilling Intelligence - Edition #1 | May 24, 2026

---

WillCo Digital Drilling Intelligence Report | Edition #1

May 24, 2026 | willcodrilling.com

---

This Edition at a Glance

- Halliburton, ExxonMobil, Sekal, and Noble executed the world's first fully closed-loop automated geological well placement in offshore Guyana, integrating rig automation, real-time subsurface interpretation, and automated hydraulics into a single commercial deployment.

- The SPE/IADC community is formally flagging that LLM-based AI systems produce non-physical outputs under incomplete input conditions - validating the physics-grounded hybrid methodology as an operational necessity, not a preference.

- The Drilling Data Management Systems market is tracking toward $3.71B in 2026 at 12% CAGR, signaling sustained operator capital commitment to digital infrastructure regardless of commodity cycle.

Why This Edition Matters

The gap between what major operators are deploying and what independent operators can actually trust is widening fast. Closed-loop autonomous drilling is no longer a conference presentation - it is running commercial wells offshore Guyana and in the North Sea. That acceleration forces a critical question for every drilling engineering team: as autonomous systems remove the human from the control loop, what ensures the physics is right? The answer emerging from both the SPE/IADC literature and field experience is not more data - it is physics-grounded constraint. That distinction is the defining fault line in drilling intelligence right now.

Top Story: World's First Fully Closed-Loop Automated Geological Well Placement - Offshore Guyana

In March 2026, Halliburton, ExxonMobil Guyana, Sekal, Noble, and the Wells Alliance Guyana team delivered what Halliburton confirmed as the industry's first fully automated geological well placement with complete rig automation in a deepwater commercial environment. The system combined Halliburton's LOGIX automation platform with Sekal's Drilltronics, integrating automated subsurface interpretation, real-time well placement decisions, and closed-loop hydraulics management - all executing without manual parameter intervention from the surface team.

The technical achievement here is not any single component in isolation. LOGIX and Drilltronics have both been fielded individually before. What Guyana demonstrated is full-stack integration: the subsurface model interpreting LWD data feeds directly into placement decisions, which feed directly into rig control, which adjusts WOB, RPM, and flow rate in real time - all within a single closed loop. No human confirmation handoff between layers. That level of system integration in a deepwater environment, with its latency constraints and wellbore complexity, represents a genuine engineering milestone.

For operators, the operational implication is direct: automated well placement at this fidelity compresses the geosteering decision cycle from hours to seconds. In Guyana's reservoir context, where staying within a thin productive interval determines well economics, that compression has measurable EUR impact. The question operators should be asking is not whether this works - Halliburton and ExxonMobil have answered that. The question is what happens when the subsurface model is wrong, and whether the closed-loop system has the physics-grounded constraints to recognize model failure before it affects the wellbore.

Key Developments This Week

Automated On-Bottom Drilling System - North Sea (Halliburton/Sekal/Equinor)

Earlier in 2025, the same Halliburton-Sekal partnership deployed the world's first automated on-bottom drilling system for Equinor on the Norwegian Continental Shelf. The system adjusted drilling parameters in real time without manual intervention, using AI models integrated across LOGIX and Drilltronics with the rig's automation control infrastructure. The North Sea deployment preceded Guyana chronologically and represents the on-bottom drilling complement to what Guyana demonstrated for well placement - together, they constitute a functionally autonomous well construction workflow across two of the world's most technically demanding offshore environments.

SPE/IADC Paper: Real-Time ROP Optimization with Physics-Safeguarded ML

Presented at the SPE/IADC International Drilling Conference in Stavanger (March 2025), paper DOI 10.2118/223713-MS details a system that uses machine learning to maximize ROP while embedding preventive safeguards against hole cleaning failures and stick-slip. The architecture is significant: ML drives the optimization, but physical constraint models govern the boundaries. When the ML recommendation would push cuttings transport beyond safe limits or into a stick-slip risk regime, the physics layer overrides it. This is the engineering pattern that separates deployable drilling AI from laboratory demonstrations - and it is showing up in peer-reviewed SPE literature with increasing specificity.

AI Framework for Unconventional Reservoirs - Middle East (IJERT, December 2025)

Research published in the International Journal of Engineering Research & Technology presents a full AI framework targeting ROP improvement, drilling dysfunction reduction, and well construction efficiency for unconventional resources in the Middle East. The framework addresses a geography where unconventional development is accelerating and where drilling dysfunction costs are disproportionately high relative to well economics. The paper reinforces that AI-driven drilling optimization is expanding beyond North American shale - operators in the Middle East, North Sea, and South America are building the same capability infrastructure.

Drilling Data Management Systems Market - $3.71B in 2026 at 12% CAGR

Market data from Coherent Market Insights places the Drilling Data Management Systems segment at $3.71B in 2026, growing at 12% CAGR within a broader digital oilfield market approaching $48B. These are not aspirational projections - they reflect committed operator capital. For drilling engineering leaders, the practical signal is straightforward: the infrastructure investment cycle is underway, and operators who delay WITSML data architecture decisions and real-time analytics platform commitments are falling behind on the foundation that all of this automation depends on.

Technology on WillCo's Radar

Large Language Models Applied Directly to Drilling Parameter Decisions

LLMs are appearing in drilling technology vendor pitches with increasing frequency, and the SPE/IADC community is right to flag the risk. LLMs trained on operational text data can generate plausible-sounding drilling recommendations that violate basic physics when inputs are incomplete or out-of-distribution - which describes most real drilling environments at some point in every well. The signal to watch: peer-reviewed SPE papers demonstrating LLM outputs validated against physics-based simulators across a full range of input conditions, with quantified failure modes. Until that body of literature exists, LLMs belong in drilling intelligence workflows as reasoning and documentation tools - not as control-loop decision engines.

WillCo Perspective: What the Engineering Says

The Guyana and North Sea deployments are genuinely impressive, and I don't minimize what Halliburton, Sekal, ExxonMobil, and Equinor have achieved. But here is what those systems require to function safely: an accurate formation model, a calibrated hydraulics model, and a wellbore stability model that reflects actual in-situ conditions. When any of those underlying physics models drift from reality - and they do, on every well, at some point - the automation layer needs hard constraints that stop it from optimizing into a problem. That's not a software feature. That's engineering judgment encoded into the system architecture before the well spud.

What pure data-science approaches miss is that drilling physics doesn't negotiate. A cuttings transport model that says you're fine doesn't become right because the ML confidence score is 94%. If the flow rate is insufficient for the hole angle and mud weight, you will pack off. The hybrid approach exists because the physics is always right and the data is sometimes wrong. That inversion - trusting the equations and validating the data - is what I want operators to internalize when they evaluate vendor automation platforms.

The question I'd put to any vendor selling autonomous drilling capability: show me the failure mode documentation. What does your system do when the formation model is 20% off on pore pressure? What does it do when torque and drag inputs are stale? If the answer is "the ML adapts," that's not an answer. That's a risk transfer to the operator.

Recommended Actions for Operators

1. Audit your WITSML data infrastructure before evaluating automation platforms. Closed-loop systems are only as good as the data quality feeding them. A data architecture assessment is the prerequisite, not the afterthought.

2. Request physics-constraint documentation from every AI/ML drilling vendor. Specifically: what physical boundaries are enforced, how are they calibrated, and what is the fallback behavior when sensor inputs are degraded or missing.

3. Evaluate the SPE/IADC safeguarded ROP framework (DOI 10.2118/223713-MS) as a reference architecture for your own analytics platform development - it represents current best practice for physics-constrained ML in a production drilling context.

4. Do not conflate RTOC capability with automation readiness. Real-time operations centers are the human-in-the-loop layer that makes automation safe. Staff and infrastructure decisions for your RTOC should precede, not follow, automation platform procurement.

5. Define your closed-loop readiness criteria now. Autonomous drilling will reach your operating environment within 18–36 months. Operators who establish internal KPIs, data standards, and physics-model calibration protocols today will capture the efficiency gains. Operators who wait will be handed a vendor's default configuration.

What We're Watching Next

- Whether Halliburton and Sekal publish quantified NPT reduction and well cost data from the Guyana deployment

- How independent operators respond to the physics-grounded constraint framework in the SPE/IADC literature

- Whether the Drilling Data Management market growth translates into WITSML architecture investment or just dashboard procurement

- William B. Contreras | WillCo Drilling Consulting | willcodrilling.com