Susan Nash is
AAPG’s Director of Innovation, Emerging Science and Technology, and a former
President of the AAPG. She has conducted and moderated many AAPG Academy
webinars, for which I am grateful, having attended many. She always asks
important questions and fosters a learning environment. Her recent article in
AAPG Explorer: The Rise of the Geoscientists’ Super Mind: Agentic AI
transforms the subsurface, is both fascinating and important. It reveals
several areas where agentic AI is making a difference in subsurface evaluation.
There are vast quantities of subsurface data in the form of seismic data, xyz
coordinates for mapping many different parameters, and other subsurface
methods.
The oil & gas industry
has been utilizing digital twins for some time. Digital twins are simply
digital models of a project, system, or process. Digital twins enable
simulations, integration, testing, monitoring, and maintenance. Wikipedia
defines a digital twin as follows:
“A digital twin is "a set of adaptive models that
emulate the behaviour of a physical system in a virtual system getting real
time data to update itself along its life cycle. The digital twin replicates
the physical system to predict failures and opportunities for changing, to
prescribe real time actions for optimizing and/or mitigating unexpected events
observing and evaluating the operating profile of the system."
AI agents are autonomous AI
entities that analyze data and make decisions based on that analysis within the
parameters set for them. Susan Nash describes the results as human intellect
being amplified by machine intelligence, creating a “Geoscientist’s Super
Mind.” She notes that the oil, gas, and geothermal industries are employing
AI agents, or agentic AI, to successfully solve problems and optimize
processes. She describes agentic AI as:
“…autonomous, goal-oriented systems capable of
independent decision-making, learning and interaction within complex
environments.”
She refers to agentic AI as a
“cognitive assistant,” consisting of software entities which are:
“…designed to perceive their environment through sensors,
process information and act upon that environment through effectors, all while
striving to achieve specific goals.”
These agents are “programmed
to execute complex workflows.” They can analyze the vast amounts of
subsurface data generated in a fraction of the time humans can and often find
hidden relationships in that data. Agents are used to update the digital twin
and to recommend actions.
“When multiple agents collaborate, they form multi-agent
systems, mirroring how a team of geoscientists distributes specialized tasks,
only at lightning speed. This intelligence adapts over time using reinforcement
learning, which teaches the agent to make optimal decisions based on past
successes and failures.”
She explains that agentic AI
utilizes frameworks consisting of multiple cloud platforms and specialized
software. She mentions the ones available from Amazon Web Services, such as
scalable compute (EC2), vast storage (S3), and a suite of AI/ML services
(SageMaker, Lookout for Equipment). Petabytes of seismic, well log, and well
production data can be housed in “data lakes,” which can be queried and
analyzed by the agents. She also mentions Haliburton’s Landmark Suite and SLB’s
DELFI platform, which are also employing AI agents. She mentions five other
platforms designed to enable geoscientists to train AI agents. i2K Connect
enables fast data processing. Bluware’s F3 platform is used for data
visualization, which enables training of agents in seismic interpretation.
Stratum Reservoirs trains agents in rock characterization and petrophysics.
Petrabytes’ platform provides data management and analysis. Quick Suite
provides a platform for automated repetitive tasks that the agents perform.
This all sounds like a multi-pronged cocktail of software suites. However, as a
geoscientist, I know well the high costs of software, subscriptions, and cloud
storage. I suspect that this type of deep-level AI analysis will likely be
confined to larger companies with more disposable income or higher budgets for
it.
Nash also notes that
companies that do energy market analysis, like Enverus and S&P Global, are
utilizing agentic AI for market forecasting, competitor analysis, and M&A
due diligence. She notes AI agent successes in enhancing operational efficiency
and de-risking investments. Below, she emphasizes their use in production
forecasting and drilling hazard mitigation:
“For production forecasting, agents continuously monitor
field performance, providing dynamic, highly accurate forecasts that allow
geoscientists to optimize capital allocation and proactively manage asset
value. This extends to economic scenario planning, where agents simulate
thousands of market conditions to ensure the technical plan is also a
financially optimized strategy.”
“Finally, in drilling hazard prediction, agents analyze
real-time MWD/LWD data and geological models to predict potential issues like
abnormally pressured zones before they occur, saving millions and enhancing
safety.”
AI agents are being used to
optimize well planning, autonomous drilling and geosteering, reservoir
management, and in geothermal development, optimizing injection and production
rates and predicting issues like scaling and corrosion.
Thus, we can see that agentic
AI is already being deployed successfully in a multitude of ways for subsurface
analysis, production, and market analysis.
“Agentic AI is the ultimate enabler, allowing geologists
and engineers to move past the data deluge and focus on the highest-value, most
complex challenges, truly becoming the architects of the subsurface energy
transition.”
Finally, she plugs an
upcoming AAPG two-day conference in Houston in April 2026 -- AI and Machine
Learning in Subsurface Energy.
Below, Nash provides some
hypothetical case studies or use cases for agentic AI applications for three
different goals: 1) optimizing infill drilling in a mature field, 2) predictive
maintenance for a geothermal binary cycle power plant, and 3) rapid due
diligence for a geothermal portfolio in considering a merger/acquisition.
References:
The
Rise of the Geoscientists’ Super Mind: Agentic AI transforms the subsurface. Susan
Nash. AAPG Explorer. December 1, 2025. The
Rise of the Geoscientists’ Super Mind: Agentic AI transforms the surbsurface
Digital
twin. Wikipedia. Digital twin -
Wikipedia
No comments:
Post a Comment