The news: Goldman Sachs is working with Anthropic to build AI agents that will automate tasks and be “digital coworkers.” The purpose is to expand employees’ speed and capabilities, according to Goldman's CIO.

How it works: The key for Goldman’s AI agents is to help employees more quickly onboard clients and complete trade reconciliations and other accounting tasks. The bank may also develop agents for monitoring employees or creating pitch decks. Per Goldman’s CIO, it's too early to tell what that might mean for headcount reductions. This has lately been a palpable concern as AI is blamed for recent and predicted layoffs.

Zoom out: The CIO added that Goldman may fire vendors as “the technology matures” and AI agents do more work. Anthropic launched several products last week, including plugins for Claude, its genAI chatbot, for contracts, finance, and customer service. The crux of these new products is their ability to autonomously perform increasingly complex work: The release sparked a panic about the future of companies in the software, legal services, and financial data industries.

Implications for banks: Goldman’s work on AI agents and the panic that followed Anthropic's product release are part of a much bigger AI trend in banking—whose adoption or development of AI tools could replace the same kinds of service providers that Anthropic’s products might. To accomplish something with AI agents, banks will need strong AI governance; a modern data layer; and for front-end applications, a framework for customer trust.

Banks should keep in mind that superficial use of AI will not be a competitive advantage in the long run. Even megabanks that are spending billions of dollars on AI and supporting infrastructure face fundamental barriers to adoption. Now is the time for board-level discussions about AI-enabled applications that could specifically influence bank operations and how to take advantage—before the technology gets further ahead of the bank.