Goldman Sachs’ tech chief says tracking individual AI usage is ineffective, instead he monitors speed from idea to production for his 12,000 engineers

(SeaPRwire) –   Goldman Sachs’ chief information officer, Marco Argenti, believes tracking each employee’s individual use of AI is not an effective way to assess whether the technology is boosting productivity.

As companies increasingly encourage staff to embrace AI in hopes of enhancing efficiency, Argenti has adopted a different strategy: he measures how swiftly his engineering teams can progress from conceiving an idea to implementing it.

Although Goldman Sachs has access to detailed data on every employee’s AI usage—including that of the 12,000 engineers under Argenti’s supervision—he told Business Insider that focusing solely on those metrics is unhelpful.

“It would be like watching only one player on the field,” Argenti explained. “Sure, this player is making more moves, but why aren’t we scoring more goals? The issue isn’t just movement—it’s about passing the ball effectively.”

A better indicator, he said, is measuring how quickly a team can develop and deploy a feature, which becomes evident when a high-performing team’s workload backlog begins to shrink rapidly.

He added that AI tools have enabled bank employees to transition from preparing PowerPoint presentations for their ideas to building prototypes that can be refined in real time based on feedback.

“There’s virtually no gap between idea and prototype now. It’s almost like ‘3D printing’ software,” Argenti remarked.

Goldman’s approach to AI

Goldman Sachs has long been a pioneer in AI adoption. In 2024, the firm launched its GS AI Platform, integrating large language models such as those from OpenAI and Google, but with added security layers to safeguard proprietary data. The company also introduced an internal version of ChatGPT and developed “Legend,” a system that streamlines research by allowing employees to search the bank’s extensive archives using natural language—without needing to know specific file names or locations.

Many organizations, particularly in the tech sector, have begun strongly advocating for AI tool adoption among employees. According to the Wall Street Journal, Google has started assessing certain software engineers based partly on their AI usage, though managers retain discretion over whether to factor this into evaluations, a spokesperson confirmed to the publication. Accenture CEO Julie Sweet recently stated that AI fluency is now essential for career advancement. Sweet has also announced that the consulting firm is “phasing out” employees who refuse or are unable to upskill in AI.

While workers at Goldman—like most professionals—have historically harbored fears or skepticism toward AI, Argenti noted a shift in attitudes among staff.

“The prevailing sentiment now is one of empowerment; people feel genuinely liberated,” he said. “A few months ago, there was significant doubt and anxiety, which I attribute largely to those who weren’t actively engaging with AI.”