IBS Intelligence Podcasts

Chandrasekhar Somasekhar
Chief Technology Officer, Cleareye.ai

Introduction

The business of trade finance is traditionally exhaustively documented. However, when large language models (LLMs) are introduced, this complexity can become significantly easier to manage for institutions responsible for reviewing and validating trade documentation.

Chandrasekhar Somasekhar is the Chief Technology Officer at Cleareye.ai, where he also leads the company’s Product Engineering team. In this interview with Robin Amler of IBS Intelligence, he discusses the adoption of LLMs in trade finance, key business use cases, and how banks can leverage AI to improve efficiency, compliance, and risk management.

Interview

Host: Robin Amler , IBS Intelligence

Introduction

Robin Amler (IBS Intelligence): I’m Robin Amler of IBS Intelligence, and you’re listening to the IBSiViews podcast. Our topic today is artificial intelligence, specifically large language models, AI programs designed for natural language processing.

I’m joined by Chandrasekhar Somasekhar, Chief Technology Officer at Cleareye.ai, who also leads the company’s product engineering team. Chandrasekhar, it’s good to meet you.

Chandrasekhar Somasekhar: Thank you, Robin. It’s good to be here.

Why LLMs Are Relevant to Trade Finance

Robin Amler: We keep hearing about large language models transforming industries, but trade finance is traditionally seen as document-heavy. Why are LLMs suddenly becoming so relevant here?

Chandrasekhar Somasekhar: Trade finance has always been document-driven, which is why large language models fit so well in this domain. LLMs can read, interpret, and extract structured information from documents much faster than humans.

They also bring contextual understanding, reading clauses, understanding intent, comparing semantically similar content across documents, and identifying discrepancies. Trade finance itself hasn’t changed, but we now have technology that can handle its complexity far more effectively.

Focusing on Business Problems

Robin Amler: There’s often a push to chase new technology. When banks explore LLMs, what does the right mindset shift look like?

Chandrasekhar Somasekhar: Every technology goes through a hype cycle, but technology should always follow the business case, not the other way around. Just because a powerful technology exists doesn’t mean it should be applied everywhere.

Banks should ask where time is being wasted, where efficiency can be improved, and where risk can be reduced. In trade finance, these pain points typically involve manual document reviews and regulatory checks. Starting there, testing the impact, and scaling gradually is far more effective than adopting technology for its own sake.

Practical Use Cases of LLMs in Trade Finance

Robin Amler: LLMs are already making an impact. Can you share practical use cases where they’re delivering results?

Chandrasekhar Somasekhar: Trade finance involves scanned documents and images flowing between banks, buyers, and sellers worldwide. Much of this data is unstructured and needs to be extracted and validated.

LLMs can read these documents in a human-like way, extract key data points, such as applicant details and shipment information, and automatically cross-check them across documents. This enables document digitisation and consistency checks, significantly reducing manual effort.

Another important use case is compliance and risk review, such as identifying red flags or onerous clauses in guarantee texts.

Fastest Impact Area

Robin Amler: Which use case delivers the fastest and most visible impact?

Chandrasekhar Somasekhar: Document digitisation and extraction validation deliver the most immediate results. The efficiency and accuracy gains are visible very quickly.

For example, LLMs can read letters of credit, invoices, or trade messages, extract beneficiary details and amounts, and cross-check consistency across documents. These tasks are traditionally manual and error-prone.

Compliance, Sanctions Screening & Risk

Robin Amler: How do LLMs help in sanctions screening and trade-based money laundering compared to traditional systems?

Chandrasekhar Somasekhar: Traditional systems rely on keyword matching, which lacks contextual understanding. LLMs understand semantics and context. They can extract relevant entities, interpret relationships, and identify misinterpretations or unacceptable clauses.

This shift from keyword-based matching to meaning-based analysis makes LLMs far more effective for compliance and risk workflows.

On-Premise vs Cloud Deployment

Robin Amler: What trade-offs should banks consider when choosing between on-premise and cloud LLMs?

Chandrasekhar Somasekhar: The main trade-offs are control versus speed, cost versus flexibility, and governance versus innovation. Cloud models allow faster experimentation, while on-premise deployments offer greater control and customisation.

Many banks adopt hybrid approaches, testing in the cloud and moving to secure environments for production. The right choice depends on a bank’s risk appetite and governance requirements.

Timelines and Benefits

Robin Amler: How quickly can banks expect to see benefits such as efficiency and accuracy?

Chandrasekhar Somasekhar: It depends on the maturity of a bank’s digitised workflows. Banks with existing digital processes can see tangible benefits within six to twelve months. Full-scale transformation across multiple document types may take eighteen to twenty-four months.

Starting small, measuring impact, and keeping humans in the loop are critical.

Human Oversight in AI Adoption

Robin Amler: Which challenge is most underrated, data privacy, model training, or human oversight?

Chandrasekhar Somasekhar: Human oversight is often underappreciated. Trade finance requires domain experts to validate results, manage exceptions, and ensure accountability. Without proper human-in-the-loop design, auditability and governance risks increase.

Human feedback is also essential for model retraining and fine-tuning. Without it, model value can decline over time.

What’s Next for LLMs in Trade Finance

Robin Amler: How do you see LLMs evolving beyond current use cases?

Chandrasekhar Somasekhar: We’re moving from automation to decision support. LLMs will connect data across documents and transactions, provide recommendations, and help predict risks.

Over time, human–AI collaboration will become more seamless, with LLMs embedded directly into trade finance workflows as decision-support copilots.

Closing

Robin Amler: Chandrasekhar Somasekhar, Chief Technology Officer at Cleareye.ai, thank you very much.

Chandrasekhar Somasekhar: Thank you, Robin.

Explore More

To learn how AI and large language models are shaping trade finance workflows, compliance, and risk management, visit Cleareye.ai.