Over the past six months, Anthropic engineers have been embedded within Goldman Sachs to design and build generative AI agents specifically for process-heavy back-office functions. These systems are intended to handle time-intensive tasks such as invoice processing, transaction reconciliation and regulatory compliance checks.

Goldman Sachs plans a phased rollout of the autonomous AI tools across its accounting and compliance divisions, integrating the models into existing workflows to minimize disruption. Initial trials focus on high-volume repetitive tasks to validate accuracy and reliability.

The bank expects the AI-driven automation to significantly reduce manual processing times and error rates, enabling faster closing cycles and improved risk management. Efficiency gains are projected to drive cost savings and support scalable operations.

Goldman Sachs has emphasized that the initiative is aimed at augmenting staff capabilities rather than triggering immediate job cuts. However, the firm intends to use productivity improvements to constrain future headcount expansion in affected departments.