Goldman Sachs builds Claude-based agents with Anthropic for back-office work

In a CNBC interview reported on February 6, 2026, Goldman's chief information officer Marco Argenti said the bank has spent about six months working with embedded Anthropic engineers on autonomous agents built on Claude. The first targets are trade and transaction accounting and client verification and onboarding. Goldman described the work as early stage with no launch date, and said job losses were premature to expect.

• About six months of work with embedded Anthropic engineers, per the CIO
• First targets: trade and transaction accounting, and client verification and onboarding
• Agents are in early stages; launch is expected soon, with no date given
• Later candidates include employee monitoring and investment banking pitchbooks