Instead of focusing on headcount reduction, Goldman's CIO measures the success of developer AI tools by their ability to consistently help projects finish ahead of schedule. This provides a tangible metric for increased output and organizational capacity.
Marco Argenti states that AI has moved beyond experimentation to become a core tool for everyday work and mission-critical applications. Companies are now expected to demonstrate concrete workflows and ROI, as the technology is delivering real, measurable results.
According to Goldman's CIO, working effectively with AI agents requires skills traditionally associated with managers: the ability to clearly explain goals, delegate tasks, and supervise output. This is fundamentally changing the talent profile companies need to hire.
Goldman's CIO notes AI has dramatically reduced the cost and time to create internal applications. This is causing a strategic shift back toward building software in-house, especially for smaller tools, leading to the termination of some third-party vendor contracts.
Goldman's CIO suggests a software vendor's vulnerability to AI depends on whether its core business process will be transformed. Regulated processes like accounting are safe, while dynamic areas like the software development lifecycle are highly susceptible to disruption.
Goldman's CIO predicts that while unit cost per token will decrease, the explosion in token usage from agentic systems will make total AI compute a major corporate expense. He suggests it should be compared to personnel costs, not traditional IT spending.
To encourage creativity, Goldman uses a central 'Model Gateway' to intelligently route queries to the most cost-effective AI model. This strategy isolates users from 'token anxiety'—the fear of consuming expensive resources—and allows a central team to optimize costs without stifling innovation.
While public AI can achieve 90% of a financial analysis, Goldman's competitive advantage lies in the final 10%. This edge is built on proprietary data, unique cross-asset class insights, global human intelligence, and expertise in complex products—factors external models cannot replicate.