BAKU, Azerbaijan, December 22. Azerbaijan can
play a vital role in applying Artificial Intelligence (AI) to
energy, Anton Aristov, Managing Director and Partner at Boston
Consulting Group (BCG) said in an exclusive interview with Trend.
Aristov noted that one of the areas in which Azerbaijan has
advantaged expertise is the energy sector.
“And so, when you combine AI with deep knowledge and expertise
in energy—particularly in oil and gas and the broader energy
transition—you create a powerful market opportunity. This is where
Azerbaijan can play a vital role in advancing the application of
AI.
The second idea we had when establishing the Caspian AI
Institute was to embed into its very DNA a principle: 70% of its
resources should be dedicated to practical use cases in energy,
while the remaining 30% should be devoted to research that pushes
the boundaries of cutting-edge technologies relevant to the global
community.
The logic behind this is simple. If you focus exclusively on use
of cases relevant to a national energy company, you will quickly
end up with an institute that is, in essence, just another “AI
department,” no different from those found in many organizations.
By contrast, committing 30% of resources to research projects of
global interest allows you to stay relevant internationally and
earn the right to engage with leading institutions such as MIT,
Stanford, Berkeley, and Imperial College. It gives you a “foot in
the door.”
This is the approach that guides our selection of innovative and
forward-looking research projects—one of which is AutonomousLab.AI,
though it is certainly not the only one,” he explained.
Aristov went on to add that recently in Dubai, Azerbaijan’s
state oil company SOCAR and the Caspian AI Institute unveiled a
breakthrough: an AI-designed carbon-capture solvent that can
significantly improve emission-reduction efficiency.
“ Yet the innovation goes far beyond identifying a single
solvent. It demonstrates AI’s ability to generate entirely new
chemical compounds. The system relies on a sequence of pre-trained
models capable of understanding the existing chemistry within a
specific domain — in this pilot, carbon-capture molecules. These
interconnected models were trained on all scientifically documented
ionic liquids used for CO₂ capture. Today, science has described
the properties of roughly 4,000 such liquids.
Once trained on this dataset, the AI was tasked with designing
new candidates — and it produced around 400,000 molecules, nearly
100 times more than what is currently known to science. This
enormous reservoir of novel, AI-designed molecules enables
researchers to predict properties computationally and carefully
“cherry-pick” the most promising options — compounds that have
never been synthesized before and may offer significantly higher
CO₂ absorption capacity,” he explained.
Aristov noted that there has been identified a portfolio of
roughly 40 candidates that are supposed to capture 50% more CO2
than what's used in the industry.
“Out of that, we selected a few and synthesized them in
laboratory in Spain. And laboratory confirmed their ability to
outperform the molecules that are used in the industry. On one
hand, you have a great example of something that actually captures
the CO2 much better. But on the other hand, you have a mechanism
that can generate novel chemistry across various domains. Think of,
for example, corrosion inhibitors, lubricant additives, or other
chemical domains that are used in the industry.
You can essentially go domain by domain: train the model on
everything that is already known, generate novel molecules, select
those predicted to outperform, synthesize them, test them in the
laboratory, and then scale to production. This is the engine that
has been built at the Caspian AI Institute. It can now move
systematically — chemical domain by chemical domain — to support
energy companies. The ambition is, in a sense, to create a
‘Chemistry 2.0 for Energy Companies’ framework: a step change
across multiple chemical domains within the energy sector,” he
said.
Aristov noted that the product is very novel, it is now being
tested it on dedicated modeling tools.
“If these tests go well, we can move to the next step — engaging
the engineers and seeing whether they are willing to test it in
actual production. While inventing a new chemical compound
typically takes around ten years, we have compressed this cycle to
roughly three months. Looking ahead, we see that the next major
bottleneck is laboratory experimentation. That is why we are
working to automate these processes, aiming to establish a fully
robotic laboratory operating 24/7, synthesizing molecules and
feeding back results — whether success or failure — in real
time.
We believe that combining AI-driven generation with robotic
automation will dramatically accelerate the development of novel
materials,” he added.
Turning Azerbaijan into innovation regional
hub
“In order to turn Azerbaijan into a regional hub of innovation,
our first priority is to build a bridge to the global AI ecosystem.
We are working with Silicon Valley investors and seeking to attract
venture capital so that this project can grow independently, at
scale, and ultimately serve other energy companies. Once we succeed
and have our first pilot proven, we will be in a position to export
the knowledge created in Azerbaijan — developed here and tested on
Azerbaijani energy challenges — to the global community,” he
said.
Aristov believes the process could also work in reverse.
“Today, it is becoming increasingly difficult for international
talent to obtain work permits in the United States. In this
context, we can create local opportunities for AI specialists by
establishing an innovation hub in one of the most attractive
regions of Azerbaijan and inviting global talent to work from here.
It could become a true two-way corridor between Baku and one of the
major global AI centers. To achieve this, we need to be recognized
within the Silicon Valley community: set up several startups there,
build visibility, and establish Azerbaijan as a source of promising
innovation. In other words, create a real bridge between Baku and
Silicon Valley,” he said.
Aristov noted that Azerbaijan now stands shoulder to shoulder
with other global innovators in AI, proudly presenting its research
at the ADIPEC Tech conference.
Konstantin Polunin, Partner and Director at BCG, for his part,
pointed out that Europe is lacking the speed of changes in the
sphere of innovations, while Azerbaijan, Central Asian countries
are doing really fast moves.
“Europe could potentially learn from these fast movers, which
propose interesting solutions. I think Azerbaijan has a right to
win, just because it's faster, it's agile, and the state could do a
lot to support this entire process. One of the ideas would be to
focus on such centers of excellence and centers of innovation,
which is a hub attracting talent, capital, scale, and also
technological application possibilities. I think SOCAR’s Caspian AI
Institute, could be a prototype for this type of center,” he
said.
Aristov highlighted the fact that Azerbaijan’s liberated
territories have now the status of special economic zone: “And
given that there is a strong push in making it a green and
innovative location, I think it's difficult to name a place where
the first hub could be better organized than this one. If we allow
data scientists, data engineers to work in Azerbaijan with some
privileges of that type, I think that could be an interesting spin,
especially if we have a bridge to the global hub.”
Accelerating decarbonization with the help of AI in
refining and chemicals
Aristov pointed out that SOCAR has made a significant leap over
the past two years — not only by setting clear emission-reduction
targets but also by developing a deep understanding of where those
emissions originate.
“I believe there is another important area we can address: the
chemistry we use. Our goal is to reduce the energy required to
produce these chemical materials, making them less carbon-intensive
and more cost-efficient. This transformation can begin in
Azerbaijan and eventually extend far beyond its borders,” he
said.
Transitioning experimental AI into practice
Polunin noted that the entire path to get to commercially
viable, applicable in industry takes years.
“I think that with radical speed-up of the first stage, there's
a chance to get to commercially viable products, not only in
chemistry, but also in exports of this chemistry, and in exports of
these capabilities. It might be that Azerbaijan will not decide to
go to develop own application, but will work together with
industrial partners around the globe who would be willing to
develop pragmatic technical solution, because they have
capabilities, and scale-up of a new chemistry is a lengthy process
where you need a lot of technical competence. But funding right
partners, right network, right licensing model, putting the first
product on the shelf, would create a momentum in Azerbaijan to go
ahead and to test other capabilities and other molecules.
We have a molecule, it proves that this is viable, and there is
a longer path to scale-up to technological applications, but that's
a question of partners, and that's a question of opening up own
capacities to test this technology, and then making a commercially
viable export product worldwide. That could be the path much faster
than if Azerbaijan would decide to go for scale-up with own
capabilities,” he said.
Compression of time required for seismic
Aristov noted that one of the projects focuses on dramatically
compressing the time required for seismic work.
“We’re reducing the routine workload of subsurface engineers who
operate complex professional software. AI agents run directly over
the interface of this software — entering data, pressing buttons,
identifying mistakes, making decisions, and taking the same paths a
human expert would, but doing so relentlessly and at high speed.
With this novel interface-level approach, we are not replacing the
professional software; rather, we are eliminating the repetitive
operational tasks that humans must perform to use it,” he
explained.
Aristov emphasized that this opens a tremendous opportunity for
human specialists to focus on the most complex and intellectually
demanding aspects of their work.
“It also enables experts to transfer their knowledge into the AI
system. When a seasoned expert is preparing to retire, the question
becomes: how do we preserve decades of accumulated insight? A
well-trained AI agent can essentially inherit this expertise and
continue the work, applying the full depth of the expert’s
knowledge. And once trained, it can be replicated across multiple
workstations.”
Aristov noted that the first prototype of this project is
already complete. “Our next step is to expand it and create a
marketplace of AI agents that accelerate different steps of the
seismic processing workflow — allowing professionals around the
world to download the specific agent they need at each stage of
their process,” he said.
Methane.AI
Aristov noted that Methane.AI, the joint project of BCG and
SOCAR, is a platform that collects data, analyzes it, and
identifies the sources and volumes of methane emissions, while also
forecasting the abatement curve required to reduce them.
“The key aspect of Methane.AI is that it is designed for
companies with limited digitalization. It enables these
organizations to conduct sampled, on-the-ground measurements and
then extrapolate the findings across the rest of their
infrastructure. It is highly tailored to the realities of
post-Soviet countries, where legacy assets often remain
non-digitalized.
We have established the Caspian Methane Accelerator, essentially
a roundtable that brings several companies together. SOCAR,
KazMunayGaz, and Uzbekneftegaz have already confirmed
participation, with more expected to join. The goal is to work
collectively on methane reduction now that Methane.AI platform give
the right framework and tool-set.”
“We use the Accelerator to build a common baseline —
understanding where methane is emitted and in what quantities — and
to jointly determine what types of measurements are needed, as well
as the level of investment required to replace parts of the
infrastructure. Investors are increasingly interested in this kind
of coordinated effort. For local Caspian companies, this
Accelerator may become a powerful mechanism to jointly lower the
cost of reducing methane emissions,” he concluded.
