Goldman Sachs has worked with artificial intelligence startup Anthropic to create AI agents to automate a growing number of roles within the bank, the company’s chief technology officer told CNBC exclusively.
The bank has worked over the past six months with Anthropic embedded engineers to co-develop autonomous agents in at least two specific areas: trade and transaction accounting, and client verification and onboarding, according to Marco Argenti, Goldman’s chief information officer.
The company is “in the early stages” of developing agents based on Anthropic’s Claude model that will reduce the time these critical functions take, Argenti said. He hopes to launch Agents “soon,” although he declined to provide a specific date.
“Think of it as a digital collaborator for many professions within the company that are large-scale, complex and very process-intensive,” he said.
Goldman Sachs CEO David Solomon said in October that his bank was embarking on a multiyear plan to reorganize itself around generative AI, the technology that has made waves since the arrival of OpenAI’s ChatGPT in late 2022. Even as investment banks like Goldman see increased revenue from trading and advisory businesses, the bank will seek to “limit headcount growth” as part of the overhaul, Solomon said.
The news from Goldman, one of the world’s leading investment banks, comes as model updates from Anthropic, co-founded by a former OpenAI executive, have sparked a sharp sell-off among software companies and their credit providers, as investors bet on who will be the winners and losers in the AI trade.
Goldman started last year by testing an autonomous AI coder called Devin, which is now widely available to engineers at the bank. But it quickly became clear that Anthropic’s AI model could work in other areas of banking, Argenti said.
“Claude is really good at coding,” Argenti said. “Is it because the coding is a bit special, or is it about the model’s ability to reason about complex problems, step by step, by applying logic?”
Argenti said the company was “surprised” at Claude’s ability to perform tasks beyond coding, particularly in areas like accounting and compliance that combine the need to analyze large amounts of data and documents while applying rules and judgment, he said.
Today at Goldman, “there are other areas of the business where we could expect the same level of automation and the same level of results that we’re seeing on the coding side,” he said.
The result is that with the help of development agents, clients will be onboarded faster and business reconciliation issues or other accounting issues will be resolved more quickly, Argenti said.
Goldman could then develop agents for tasks such as monitoring employees or writing pitchbooks for investment banks, he said.
While the bank employs thousands of people in compliance and accounting functions where AI agents will soon operate, Argenti said it was “premature” to expect the technology to lead to job losses for these workers.
Still, Goldman may phase out the third-party vendors it uses today as AI technology matures, he said.
“It’s always a compromise,” Argenti said. “Our current philosophy is to inject capacity, which in most cases will allow us to do things faster, resulting in a better customer experience and more business.”