AI Adoption in Commodity Trading

AI Adoption in Commodity Trading
Introduction
In this conversation, Nicholas Frank speaks with Vadim Linchevsky about where AI is delivering
value today, what is holding firms back, and how organisations should approach adoption.
Artificial intelligence is rapidly moving from concept to application across commodity trading.
From operational workflows to decision-making, firms are beginning to explore how AI can be
embedded into the fabric of their organisations. Yet adoption remains uneven. Much of the
industry still relies on fragmented systems, manual processes, and information that lives outside
structured platforms.
Vadim Linchevsky is a veteran metals trader, having worked at Gerald Metals and on Gunvor’s
early metals desk. He has seen first-hand the complexity of commodity trading operations and
systems. He has also led trade finance as a banker, overseeing the full commodity financing
origination and execution process, and has witnessed the strengths and weaknesses of various
supporting technologies and processes.
Vadim is currently engaged in the adoption of AI in commodity trading markets, building agentic AI
workspaces with conversational interfaces purpose-built for trade operations. He is the founder of
Kansofy Inc., an AI infrastructure venture helping companies move from manual processes (emails,
spreadsheets, human memory) to systems that assemble context and reduce noise.
Nick Frank is a Director at Options Group within the Commodity Trading Practice, with 25 years of
experience in Executive Search.
Nicholas Frank (Options Group) in
conversation with Vadim Linchevsky (Kansofy Inc)
Nicholas Frank
Director
Options Group
Vadim Linchevsky
Founder
Kansofy Inc
Nick Frank (Options Group): Vadim, I read with great interest your recent thesis on the adoption of
AI for commodity trading operations. In your opinion, what are the bottlenecks, inefficiencies, or
other problems in commodity trading that we can expect AI to alleviate?
Vadim Linchevsky (Kansofy Inc): Most operational reality lives outside formal systems.
Call it the “information shadow”: the context in email threads, verbal commitments, relationship
history, and exception agreements. In my research, systems capture maybe 20% of what actually
happens. The rest never makes it into structured fields.
This is where most AI implementations miss the point. The constraint is not data quality; it is data
coverage. Your E/CTRM might be clean, but it only captures a fraction of reality. The other 80%
lives in emails, documents, and chats, flows that no structured system ever touches.
I have seen it everywhere. One company I studied had 850 staff and a proper ERP system, yet still
needed three people spending their days chasing vessels by email and phone. Internal procedures
were met but coordination remained unresolved.
The real value lies in applying AI to unstructured flows, information that has always been valuable
but never accessible at scale.
Right now, your best people spend a third of their time routing information, reconciling data, and
bridging system gaps. That is not judgment, it is overhead.
AI reclaims that capacity, not by replacing people, but by handling information work that was
never supposed to be theirs.
Nick Frank (Options Group): How do firms that are used to structured, manual processes
transition to trusting automation and AI?
Vadim Linchevsky (Kansofy Inc): Trust is built on transparency and consistency. That does not
change with AI, what changes is how you achieve it.
For AI, transparency comes down to three things:
Auditability: A full trace from output to input. If the AI flags something, you can see why, what
data, what logic, what source. Not just what it concluded, but how it got there.
Observability: Not just logging for compliance, but visibility for operators. They can see what
the AI saw before it reached a conclusion. No black boxes.
Simplicity: Binary decisions where possible, with clear boundaries between what the AI decides
and what it surfaces for human judgment. Complexity is where hallucinations hide, the simpler
the decision space, the easier it is to verify.
The practical path is incremental. Start in read-only mode. Let the system observe, classify, and
surface patterns, do not automate anything yet. Trust is earned through demonstrated value, not
promised capability.
There’s also organisational memory to contend with. One practitioner told me: “Blockchain came
in 15 years ago, and for 15 years people have been saying it’s going to solve problems.” That
scepticism is earned. You overcome it through demonstrated results, not promised
transformation.
When people can see how it works, trust follows.
Nick Frank (Options Group): What are your thoughts on AI being another ‘megatrend’ or bubble?
Vadim Linchevsky (Kansofy Inc): The scepticism is understandable. $660 billion in AI spending by
tech companies this year alone is mind-bending, no wonder investors are getting uneasy. But I also
know what I am looking at.
Enterprise AI reached $37 billion in 2025, 3.2x growth in a single year, making it the fastest-scaling
software category ever measured. 78% of organisations now use AI in at least one function.
Goldman Sachs reports AI completing 95% of an IPO prospectus in minutes, work that used to take
two weeks and six people. Anthropic’s latest research projects 1.0–1.8 percentage points of annual
productivity growth.
Last year, McKinsey reported that commodity traders have been slower than other industries to
digitalise, with $5 billion in unrealised efficiencies still untapped. Why? Data is walled off in silos,
trading desks, logistics, and operations all working from separate, incomplete views.
That is not theory. We automated customer briefs for a mid-market client from one hour down to
six minutes, but only after spending two months assembling the context first.
That is the gap, not capability, but infrastructure. AI can’t improve what it can’t see. And most
operational reality still lives in spreadsheets, email threads, and side conversations, the
“information shadow.” The companies pulling ahead are not the biggest spenders; they are the
ones that made their context accessible first.
There is a bubble in expectations, not in the technology. Jacquard’s punch cards automated
complex patterning in 1804. Two centuries later, the constraint is the same: you can only process
what you can access.
Nick Frank (Options Group): Does the rise of AI agents mean the replacement of humans in some
job roles?
Vadim Linchevsky (Kansofy Inc): Amplification, not displacement, that’s the pattern that works.
The people I have worked with are not being replaced by AI; they are being freed from work that
was never supposed to be theirs, routing information, reconciling spreadsheets, searching for
documents. This isn’t judgment; it’s overhead.
My research made the preference clear: practitioners want to extend judgment, not delegate it.
One said, “I want AI to be my policeman, to tap me on the shoulder and say, ‘watch out for that
one.’” Another said, “I would automate as much as possible, but I want the human to send
reminders to the customer.”
In many organisations, your best people do not follow the process, they are the process.
Institutional knowledge, relationship context, and exception handling live in their heads because
no system ever captured them.
That is not sustainable.
When you reduce overhead, you do not eliminate roles, you change what those roles can do. Time
that went to reconciliation goes to exception handling. Time spent being the process goes to
improving it.
The risk is not AI taking jobs, its organisations optimising for headcount reduction instead of
capability expansion.
What can your people do when they stop being the glue?
Nick Frank (Options Group): In summary, what is your advice to companies in the early stages of AI
adoption?
Vadim Linchevsky (Kansofy Inc): The tools are moving fast. Capabilities that required an
engineering team a year ago are now accessible to a mid-market company with one technical
person and a clear problem. The window is wider than most people realise.
But there is a catch my research kept surfacing; I call it the “cognitive bandwidth paradox.” Loss
aversion, cognitive overload, and status quo bias reinforce each other. When your people are
saturated, they cannot evaluate alternatives. The organisations that would benefit most from
efficiency gains are often too overwhelmed to pursue them. One interviewee put it plainly: “Not
doing AI now is like ignoring the internet.” He’s right, but the people who need to hear that are
buried in 400 emails a day.
So don’t start with automation, start with observation. Pick a process that is repetitive, highvolume,
and low risk. Let a system watch, classify, and surface patterns, but don’t let it act yet. The
goal isn’t efficiency; it’s visibility. Most organisations do not know where their information flow
breaks down, they just feel the friction. AI makes the invisible visible. Once you can see it, you can
fix it.
The talent bottleneck is real. Implementations that stall are those where no one in the room
speaks both languages, technology, and physical trade. Find that person: hire them, partner with
them, or develop them internally. That translation layer is where everything connects or doesn’t.
But here is what I would say to anyone running a trading desk or financing book today: the
constraint was never technology or budget. It’s that your best people are spending a third of their
time moving information instead of using it. That’s the number to fix, everything else follows.
For more information or to discuss any of the topics above please feel free to contact Nick
Frank: nfrank@optionsgroup.com
Options Group
Options Group is a leading global executive search and strategic consulting firm specialising in
financial services, including energy/commodities, capital markets, global markets, alternative
investments, hedge funds, and private banking/wealth management.
For more than 30 years, our clients and candidates have come to know that Options Group
consistently exceeds expectations. Our innovative approach to facilitating search solutions for
clients and candidates worldwide has become an industry standard.
With a global footprint of 14 offices across five continents, Options Group has the resources,
analytics, and experience to deliver top-tier talent to firms of all sizes. We are at the forefront of a
new generation of human capital solutions, where research meets recruiting to drive optimal
strategic outcomes. Options Group takes a hands-on approach in providing a superior range of
solutions for financial services and technology clients and candidates.