Goldman Sachs is collaborating with artificial intelligence company Anthropic to develop AI agents capable of performing core banking functions, including trade reconciliation, accounting, and client onboarding. Engineers from Anthropic have been embedded within Goldman Sachs for six months, co-developing software with the bank's technology team using Anthropic's Claude model.
Marco Argenti, Goldman Sachs' Chief Information Officer, stated the AI agents are currently in testing but will be deployed soon to automate "process intensive" tasks. "Think of it as a digital co-worker for many of the professions within the firm that are scaled, are complex and very process intensive," Argenti said. The initiative is part of CEO David Solomon's long-term strategy to integrate generative AI across the company to increase operational efficiency and reduce reliance on new hires and third-party contractors.
Concurrently, Anthropic's release of new Claude tools has triggered significant market volatility. The updates, which include features for legal work, data analysis, and autonomous workflow management, led to a multi-day selloff in stocks of software companies like Salesforce, Intuit, and Adobe. Investors reacted to the perceived threat that Claude's capabilities could displace human roles in legal, financial, and technical teams.
Anthropic's strategic focus on enterprise and developer tools has driven its rapid adoption. Data from expense software company Ramp indicated Claude accounted for nearly 80% of API usage across third-party services in January 2024. The company employs a unique "reinforcement learning from AI feedback" training method and projects reaching profitability by 2028.
