We want to be a people-first organisation: Goldman Sachs CIO Marco Argenti

Engineers need to think about the problem first and not the solution, says Argenti

Goldman Sachs, the storied American investment bank, is embracing generative AI (artificial intelligence) across its global operations, including its 9,000-strong workforce in India. In this exclusive interview with Business Standard, Chief Information Officer Marco Argenti discusses with Avik Das his strategic priorities, the evolving role of engineers, and why being people-first matters. Joining him, Gunjan Samtani, Co-chairman of Goldman Sachs in India and Country Head of Goldman Sachs Services India. Edited excerpts…

 

What are the three most important strategic priorities for you in the technology world?

 

Argenti:  They are AI, movement to the cloud, and uplifting the quality of data. We have to work synergistically around those three elements. And then we apply AI to gain efficiency and agility, and also find the ways people work. If companies need to leverage AI, they need to work on their data first because with bad data you have bad AI. We are in the middle of one of the greatest transformations with regard to the people aspect. We cannot look at this transformation with just a technology angle. Technology is only a tool and with AI, people need to develop new habits around the way they perform their tasks. Our entire engineering population has been enabled with some AI tools and there is a need to evolve the engineering profession and the way they do their work. 

 

Can you describe those changes in the engineering profession and its impact?

 

Argenti: Engineers need to think about the problem first and not the solution. We are introducing this to shift from the pattern of how, what, and why to why, what, and how, and really work backwards to understand the problem we are trying to solve before building the software. In the current world, every engineer also needs to be product manager and understand how to get to the heart of the problem. You cannot be just a pure coder or an individual contributor. An engineer needs to be on top of what his product does. When you have agents working for you, you need a little bit of managerial skill too. You have to describe what you want the AI to do, break down the task that you can delegate, and supervise the work of AI. And that changes the profession of an engineer. This is because while AI can code better than humans today, the operations cannot be left alone because mistakes can be costly and the presence of a human in the loop is absolutely fundamental.

 

Where is Goldman Sachs in terms of being an AI-first organisation bank?

 

Argenti: We want to continue to be a people-first organisation because that has always been our strength, but we believe that people will be empowered by AI. Having the best people is even more central today because they will get the best out of AI. People are a true multiplier for AI. So people-centricity will be at our core.

 

We are among the first companies in our sector to inject AI into everything we do. That includes building a GS (Goldman Sachs) AI platform and making it a foundational element of pretty much every AI application that we built. We are giving our developers AI tools early and pushing the envelope on what agentic capabilities can do for us. AI is now at the centre of our strategy and we see it as a means to empower our people and not substitute them.

 

Can your centres in Bengaluru and Hyderabad become the gen AI or agentic AI centre in the future?

 

Argenti: Broadly, we have an incredible opportunity in our India offices. Our Bengaluru and Hyderabad offices are where we have developed some deeper expertise on AI, such as how to apply AI agents across the organisation. This is also where a lot of our workflow comes together, in terms of various teams (across engineering and businesses) working as one. We have the best people here, a favourable cost structure and the best AI opportunity. Together, this is an opportunity that is probably unprecedented for not just Goldman Sachs here but for India too.  

 

Samtani: We have been in India for over two decades now and have developed front-to-back stacks for each of our businesses collectively in these offices. That means we have developed both product and business expertise. We continue to grow and there are still over 500 engineering open roles in India for AI-related skillsets. With expertise in the business domain and engineering platforms, coupled with a culture of continued learning and our impact on the India GCC (global capability centre) ecosystem, our Bengaluru and Hyderabad offices are key transformation hubs for the firm, and enable AI at scale for Goldman Sachs. 

 

How do you manage the AI agents and what are the skills an engineer needs?

 

Argenti: We need to be open-minded about these changes. I think one of the skills that are super-important is communication, which is having the ability to have an active dialogue with the business to unpack what they are trying to solve and translate that into prompt engineering. Prompt engineering is the most basic skill one should have and this is almost like learning a new language.

 

Samtani: The intersection of AI skillsets with values, such as teamwork, collaboration, and empathy, is important for engineers. It is this amalgamation that will define the evolution of high-performing organisation. As much as the skillsets of the future, client excellence and integrity will remain essential in this AI-led world.

 

What are your views on Apple’s research paper that says large reasoning models tend to collapse with increasingly complex problems?

 

Argenti: The same could be said sometimes for humans too. We make mistakes and can be overwhelmed by very complex problems but that does not mean we cannot still do excellent work. So it is like being conscious that those large AI reasoning models are not the solution to everything. However, there is plenty of evidence that when you define the scope of the problem and describe it, AI can do an incredibly good job. If you take something to the extreme, you will always find that problem.