Making AI agents smart isn't the hard part — it's teaching them company culture, says a Goldman Sachs exec
Goldman Sachs; Chelsea Jia Feng/BI
- Goldman Sachs' Marco Argenti said businesses need to integrate company culture into AI agents.
- The tech industry has mostly focused on the "specialization of expertise," he said.
- AI agents — AI that acts independently — are a hot topic in tech.
Goldman Sachs' chief information officer said the real challenge of using AI agents in the workplace isn't just making them smarter — it's teaching them company culture.
Marco Argenti said on an episode of the Goldman Sachs Exchanges podcast published Tuesday that businesses must figure out how to "inject" their cultural traits and leadership principles into AI agents, just as they do with human employees.
The tech industry has mostly focused on the "specialization of expertise," he said. But it doesn't talk much about the "enhancement of the cultural aspect of agents," which he said is crucial for managing AI in the workplace.
AI agents — AI that can act and make decisions independently — are a hot topic in tech and the workforce. Nvidia CEO Jensen Huang said in January that "the age of agentic AI is here" and said last year that Nvidia's 50,000-person staff could work with 100 million agents. OpenAI's Sam Altman predicted that AI agents could start entering the workforce this year.
AI agents are intended to be digital coworkers or assistants to human workers in fields ranging from healthcare and supply chain management to cybersecurity and customer service.
Argenti warned that without cultural training, these agents risk missing the nuances that define an organization's identity.
Comparing AI agents to new hires, Argenti said they might become more intelligent over time but won't automatically become culturally smarter.
Businesses must figure out how to embed their unique identity into AI agents so they can interpret work through the company's lens and communicate in its language.
For instance, he said, an AI agent summarizing portfolio analysis in private wealth management at Goldman Sachs needs to apply a "Goldman lens" to its answers. That means translating financial content in a way that aligns with the firm's approach.
"Cultures are very peculiar," Argenti added. "They're hard to describe, but you see them when you see them."
Argenti acknowledged that he does not know what culturally trained agents look like in practice, asking, "What are the actual leadership principles or tenets that an effective corporate agent should adhere to in order to fit into the culture of an organization?"
His answer: "That's a very complex problem that I don't think has been solved yet."