Goldman Sachs Partners with Anthropic to Build AI Agents for Back-Office Automation

Goldman Sachs Goes All-In on AI Agents

Goldman Sachs is taking a major leap into autonomous AI by partnering with Anthropic, the company behind the Claude AI model, to build AI agents capable of handling complex back-office operations. Chief Information Officer Marco Argenti revealed that embedded Anthropic engineers have spent the past six months working side by side with Goldman staff to develop these systems.

What the Agents Will Do

The AI agents target two primary areas: accounting for trades and transactions, and client vetting and onboarding. These are traditionally labour-intensive roles requiring staff to analyse large volumes of data under strict compliance requirements. Argenti described the agents as digital co-workers for professions within the firm that are scaled, complex, and process-intensive.

A Surprise Beyond Coding

Goldman executives were reportedly surprised by how capable Claude proved at tasks beyond coding. The model showed particular strength in areas like accounting and compliance, which demand the ability to parse large datasets while applying rules and judgement. The firm now sees potential for the same level of automation across multiple departments, with future applications possibly including employee surveillance and investment banking pitchbooks.

The Broader Wall Street AI Race

Goldman is not alone in this push. Wall Street banks are collectively planning to spend eighty-five billion dollars on AI in twenty twenty-six. JPMorgan Chase has moved from pilot projects to over four hundred production use cases and committed eighteen billion dollars to technology investments with over two thousand AI specialists. Bank of America reports ninety percent workforce AI adoption across its operations.

Anthropic's Enterprise Momentum

Anthropic, now valued at roughly three hundred and fifty billion dollars after its latest funding round, has expanded its enterprise client base from fewer than one thousand businesses to over three hundred thousand in two years. The company's annualised revenue run rate has surpassed nine billion dollars, with enterprise customers accounting for eighty percent of revenue.