Goldman Sachs is making strides in using artificial intelligence to streamline its back-office operations. The bank plans to implement Anthropic’s Claude model to assist with trade accounting and client onboarding tasks. This move is part of a larger trend among big banks to leverage generative AI for improving efficiency and reducing manual work.

Enhancing Back-Office Processes with AI

Traditionally, operational processes like document review, reconciliation, and compliance checks have relied on large teams performing repetitive tasks. While some automation has been used in the past, Goldman Sachs sees AI, especially generative models, as a way to handle more complex cases that rules-based software struggles with. According to Marco Argenti, the bank’s chief information officer, most transactions fall within clear rules, but a small percentage involve edge cases requiring judgment.

For example, in KYC (know-your-customer) procedures, minor discrepancies or nearing-expiry documents create situations where human intervention is usually needed. Argenti explains that neural networks can analyze these micro-decisions by applying contextual reasoning, something fixed rules systems can’t do. This means AI can help manage exceptions more efficiently without replacing existing rules, mainly reducing the number of cases that need manual review and speeding up resolutions.

Using AI to Improve Coding and Development Workflows

Goldman Sachs has already experimented with Anthropic’s Claude models internally, especially in software development. Developers use a version of Claude combined with Cognition’s Devin agent to assist with programming tasks. Human developers set the goals and regulatory limits, and the AI generates code, which is then reviewed by people. The AI also helps run tests and validate code, making the development process faster and more productive.

This approach has changed how developers work, allowing them to complete projects more quickly. The AI acts as a helpful assistant, following specific instructions to produce useful outputs. Goldman Sachs believes that this not only boosts productivity but also improves the quality of the software they develop.

Applying AI to Trade and Client Onboarding

For trade accounting and onboarding new clients, Goldman Sachs and Anthropic’s AI project teams studied existing workflows with domain experts. They looked for bottlenecks and areas where AI could make a difference. The goal is to automate routine parts of these processes, freeing up staff to focus on more complex or high-value tasks.

By analyzing the current steps, the teams identified opportunities where generative AI could help with data collection, document assembly, and validation. This targeted use of AI aims to reduce manual effort, cut down processing times, and improve accuracy in critical operations like trade reconciliation and KYC compliance. Goldman Sachs sees this as a way to stay competitive by making its back-office functions more agile and less prone to human error.

Overall, Goldman Sachs’s adoption of Anthropic’s AI systems highlights a broader shift in the banking industry. Large financial institutions are increasingly turning to advanced AI tools to handle complex operational tasks, improve efficiency, and support their digital transformation efforts. This move could set a new standard for how banks leverage artificial intelligence in their everyday workflows.

Inspired by

• https://www.artificialintelligence-news.com/news/goldman-sachs-ai-deploys-anthropic-systems-with-success/

Sources

• americanbanker.com