From Where I Sit - Marco Argenti

From where I sit as Chief Information Officer at Goldman Sachs, I have the privilege of partnering with our business leads on driving innovation and transformation across the firm. As with any organization, a major focus of the past few years has been on Artificial Intelligence, and its potential to transform how we work. At a time of unprecedented change, it’s clear to me that AI isn’t just accelerating, it’s evolving in a way that feels almost biological: messy, non-linear, and full of unexpected leaps. For those who have witnessed previous waves of technological change—over my 20 plus years I’ve seen a few—the current acceleration in AI feels fundamentally different.

We are already seeing AI’s transformative power in action across the firm. For instance, we are piloting Devin, an autonomous software engineer. Devin builds a plan, writes code, and proposes updates for complex engineering tasks—such as upgrading dependencies or migrating code from one language to another—executing only after our human engineers give the green light. Acting as a force multiplier, Devin reduces developer toil and accelerates both the speed and scale of our development capabilities. And, our internal GS AI Assistant—a secure application that connects our people to leading generative AI models—now includes cutting-edge models like GPT-5, OpenAI’s most advanced LLM. By integrating new models quickly, we ensure our people have timely access to the best available capabilities and can easily tailor their AI experience to the task at hand—whether that’s summarizing documents, learning technical concepts, or generating client-ready content—freeing up valuable time for client engagement and high-value work.

The powerful AI capabilities we're deploying today are paving the way for the next frontier: the rise of agents—AIs that don’t just answer questions but roll up their sleeves and get to work—prompting us to rethink not only our workflows, but our entire approach to problem-solving.

If conversational AI is the helpful assistant at your desk, agents are the colleagues who take on entire projects—such as quietly extracting company legal names and addresses from a folder of contracts, grouping them by country or state, and updating the list every time a new PDF arrives. This kind of automation goes beyond simple Q&A, requiring the agent to break down the problem into steps, monitor changes, and execute actions over time. This evolution will drastically change team composition, productivity, and the way we think about our workforce and scale.

Of course, with new capabilities come new responsibilities. As AI systems become more capable, the ethical dimension grows ever more critical. We are, in essence, the parents of a new species—responsible for shaping its values, logic, and behavior. The study and application of ethics, education, and moral principles must now extend to AIs, with humans retaining a disproportionate responsibility as policy makers and educators. What does this mean for our 12,000 strong team of engineers at Goldman Sachs? Will it change how we hire, train and empower our people? Most certainly.

At the same time, the progressive introduction of agents in the workforce continues to pose questions about the way we integrate them with humans: how does the concept of apprenticeship—arguably one of our firm’s keys to success—translate to AI, which has a potentially infinite career span, and will continuously learn, adapt and evolve? What new skills do we need to equip our human talent with, to successfully co-exist and thrive in the presence of AI colleagues? At a minimum, any employee interacting with agents will have to develop the ability to communicate, delegate and supervise the work of their AI counterparts—which are, in a nutshell, manager skills. In my opinion, the latter is the most crucial because of the risks that unsupervised agents may pose, and the human tendency to over-rely on answers and actions that are, by the very nature of the models, optimized for plausibility.

Throughout this journey, the evolution from models to agents—and the ethical imperatives that follow—invite us to continually reflect on our own roles, responsibilities, and readiness. Rather than saving such questions for the end, they must inform our approach at every stage. In a world where change is the only constant, perhaps the real question is not whether we’re ready, but how we’ll choose to engage—using AI not as a tool, but as a partner in progress.

As we navigate this transformative moment, I am grateful for the legacy built by our alumni and energized by the possibilities ahead. Together, we continue to shape the future of Goldman Sachs Engineering.

I may have offered more questions than answers, but I hope they spark reflection. I welcome your thoughts and feedback—please feel free to reach out at Marco.Argenti@gs.com or connect with me on LinkedIn.