AI
Exclusive: Raghav Sehgal on the Shift Towards Agentic AI in Financial Services

Raghav Sehgal, General Manager for the Middle East at Model ML, discusses how agentic AI is changing the way financial professionals approach their work and what this means for the industry.
[For more news, click here]
The growing presence of artificial intelligence in financial services is bringing a new set of questions for an industry built around complex information, high-value decisions and established institutional practices. As the technology advances, financial organizations are reassessing its potential to reshape professional services and the way their teams operate.
Across the Middle East, investment in technology and the expansion of financial markets are creating new opportunities for AI companies looking to establish relationships with banks, investment firms and institutional investors. From the UAE and Saudi Arabia to other emerging financial centers in the region, the intersection of financial services and advanced technology is becoming an increasingly important part of the industry’s development.
One company operating at this intersection is Model ML, the agentic operating system for financial services. With its expansion into the Middle East, the company is looking to build a stronger presence in a region where financial institutions are increasingly engaging with the possibilities of enterprise AI.
In this exclusive Q&A with Tech Revolt, Raghav Sehgal, General Manager for the Middle East at Model ML, discusses the evolving relationship between artificial intelligence and financial services, and the opportunities and challenges shaping the industry’s next phase.
Rethinking AI in Financial Services
1. Can you give a brief overview of what Model ML does?
Model ML is an agentic harness built for financial services firms. To explain what that means, rather than using AI to explain AI, I like to use a laptop analogy. When you sit down to work, you do not think about who made the processor, memory, keyboard or screen. You care about how those components come together to help you get the job done.
We are building the laptop of the agentic world for bankers, consultants and private equity investors. Our platform brings together AI models, agents, skills, memory, and internal and external data, organized around how each person and institution works.
That allows teams to automate work across the whole process: researching and analyzing information, turning that analysis into presentations and financial models, and reviewing those materials before they reach a client or investment committee. The aim is not to give people another chat window. It is to help them get the work done.
2. Having spent many years working in consulting prior to joining Model ML and now leading the consulting practice, can you speak to the trends you’re seeing in AI adoption across your largest consulting clients and explain why you see such a strong product-market fit?
I have lived the problem we are trying to solve. At Bain, I was the person up at two in the morning building slides and moving boxes around. That work is necessary, but it is not why most people become consultants.
What I am seeing is that AI adoption is becoming as much an organizational question as a technology question. The conversation is shifting from “What can this platform do?” to “How do we embed it into the way our teams work?” Senior leaders are taking a more active role in deployment, and firms are building networks of AI champions to share practical examples and best practices.
That is why our role goes beyond providing software. We work alongside clients as an AI-enablement partner, helping them identify where the technology can deliver value and supporting the changes needed to realize it. Working across industries and clients globally gives us the benefit of pattern recognition: we can bring what we have learned elsewhere to help each firm move more quickly from experimentation to tangible ROI.
The product-market fit is strong because we are building around the work consultants actually do: researching information, developing an argument, turning it into a presentation or model, and rigorously checking the output. Helping with that connected process frees people to spend more time on the judgment, insights and client relationships that drew them to consulting in the first place.
3. Why do you think that large firms should choose to procure a vertical application as opposed to working with the labs? Where does the advantage lie?
I would not frame it as a choice between us and the labs. What is less visible from the outside is how much cooperation takes place across the industry. We work closely with the labs, and advances in their models are an important part of what makes our product better.
The distinction customers should make is between access to a powerful model and a product built to deliver outcomes. To return to the laptop analogy, a great processor matters enormously, but it is not the whole computer. Our role is to bring that intelligence together with the right data, models, tools and instructions for financial work.
For a financial institution, that means connecting its information, applying its ways of working and producing useful deliverables. The advantage lies in how those pieces work together, not simply in which model sits underneath them.
Our approach is also deliberately model-agnostic. The labs have built a harness centered on one model. We’re building a harness that brings together frontier models and lower-cost, open-weight models fine-tuned for finance. We route tasks to the appropriate model and level of reasoning based on what the work requires. That allows us to balance quality, speed and cost while keeping the client focused on the outcome, rather than the technology needed to produce it.
4. Model ML recently opened its Dubai office to work more closely with financial institutions across the Gulf. What are banks and sovereign investors in the region asking for that differs from the needs you see in other markets?
I would be careful about drawing too sharp a distinction. The underlying ambition is familiar: help people make better use of information and spend less time on manual work. Comparing our Gulf conversations with those in the rest of the world, the priorities I would highlight are greater control over sensitive data, fitting into the institution’s own investment processes, and hands-on support with adoption.
For sovereign investors in particular, questions about where sensitive information can be stored and processed are central to the conversation. Teams also want AI to follow their established ways of assessing investments, not simply produce a plausible-looking answer.
There is also a very practical request: do not just hand us software and disappear. Work alongside our people, understand the recurring tasks, and help make the product useful in their day-to-day work. That is why being physically present matters. Sitting with a team and helping someone solve a real problem can do far more for adoption than another remote demonstration.
5. Model ML’s platform can route different tasks to different AI models. How do you decide which model is best suited to a particular financial task, and how do you balance output quality against speed and cost?
We start with the task, not the model’s brand. Our engineers evaluate models against analytical finance and consulting work, assessing both the quality of the output and the speed and cost of producing it. Those evaluations guide our model choices, and we revisit them as models evolve.
Drafting a short email is very different from producing an investment analysis that draws on several internal systems and needs to retain substantial institutional context. A simpler task may be well served by a faster, less expensive model; a more demanding task may warrant a more capable model with deeper reasoning.
The principle is not to use the most expensive model for everything, or the cheapest regardless of the result. It is to meet the task’s quality requirements as efficiently as possible. Model choice is also only one part of the equation: the data, instructions and tools available to the agent all shape the outcome.
There is no single “best” model for every task. The right choice depends on the accuracy, quality, speed and cost required for a particular outcome. Clients should be able to focus on the work they need done, rather than having to navigate those trade-offs at every step.
The Changing Role of Financial Expertise
Financial services has always depended on the ability to interpret information, assess risk and make decisions where the consequences matter. AI is beginning to change how much of the work leading up to those decisions is carried out, but it does not remove the need for professional judgment. If anything, it places greater importance on knowing which findings to trust, what deserves closer scrutiny and where experience makes the difference. That balance will be an important part of how financial services evolves as AI becomes more widely used.
About Raghav Sehgal
Raghav Sehgal is General Manager for the Middle East at Model ML. He began his career in Bain & Company’s private equity practice, working in London, New York, and San Francisco. After earning an MBA from Harvard Business School, he joined Model ML in London shortly after it emerged from stealth, as its Founding GTM hire in EMEA. He later became Head of GTM for the Consulting and Private Markets verticals. In his current role, he works closely with Model ML’s clients to help integrate the company's agentic AI platform into their operations, bridging the gap between the technology and its practical application across financial services.





