Barclays Expands Claude Across the Bank

British bank Barclays is expanding its use of Anthropic's Claude across the company. In an announcement on October 1 (local time), Anthropic said Barclays is broadening its strategic collaboration with Anthropic to integrate enterprise AI systems across its global operations. The most concrete commitment is developer adoption. Barclays plans to have 50% of its developers using Claude Code by the end of 2026, expanding to a majority of its software engineers in 2027. The original Anthropic announcement that METAL reviewed contains remarks from three executives at the two companies, along with usage figures for two deployments that are already live.

Barclays set out three goals for the expansion: speeding up software development, modernizing aging legacy systems, and improving the bank's operational efficiency. Barclays is a British universal bank. The announcement noted that bringing AI into a large, tightly regulated organization requires disciplined governance as well as capable technology, and described the expanded collaboration as a deployment aimed at real outcomes, such as faster support for customers.

The deployments already running show the scale. Colleague Knowledge Assistant, an internal tool used by Barclays UK staff, helps employees quickly find the information and answers they need when serving more than 20 million UK retail customers. Live since 2025, the tool uses Claude through a retrieval-augmented generation (RAG) architecture. When a question comes in, relevant material is first retrieved from internal sources and passed to the model, which then builds its answer from that material. More than 16,000 employees have adopted the tool so far, and it has handled over one million searches.

The second deployment is client inquiry handling in the Global Markets business. Teams there use Claude models to classify incoming emails, enrich them with the information needed, and determine the best processing route. At this stage they also check whether a request contains enough information for operations staff to act on it right away, and set priorities. The platform processes roughly 120,000 emails a day. Anthropic said the system has reduced manual handling and helped staff identify and act on relevant information more efficiently.

"The true measure of any technology is the impact it has on customers, clients, and colleagues, and the outcomes it enables," said Anne Marie Darling, Group Co-Chief Operating Officer (COO) at Barclays. "Expanding the use of Claude across Barclays is about more than adopting AI," she said. "It is about transforming how work gets done." Darling explained that by cutting the time spent on routine work and streamlining decision-making, the bank wants staff to apply their expertise to more complex problems.

The plan for the technology organization came from the other Group Co-COO, Craig Bright. "Software engineering and cyber security are both being reshaped by increasingly capable AI systems," Bright said. He explained that Barclays is moving toward AI as an agentic capability embedded in how it builds, tests, secures and operates technology. According to Bright, Barclays is using Claude to modernize legacy platforms, improve software quality, and let its technical experts focus on the most complex challenges.

Paul Smith, Chief Commercial Officer (CCO) at Anthropic, called the collaboration "a significant milestone for us in the UK." "Barclays serves millions of customers and clients across the globe, so their standard for AI is high," Smith said. "That's why Barclays is rolling Claude out with strong security and oversight." He said Claude now helps 16,000 Barclays employees find answers for customers, sorts 120,000 emails a day, and will be in the hands of most Barclays engineers by 2027. According to the announcement, Barclays applies governance, security controls and human oversight to each AI use case.

AI deployment in finance is spreading from the back office outward. According to an industry survey released in August, 95% of banks and insurers were broadly deploying or more deeply embedding new AI tools in back-office functions such as data and technology, and the same 95% said new AI had proven very or extremely effective in the functions where it is used. Anthropic has also been adding products aimed at financial customers. METAL previously reported that Anthropic launched a Claude plugin for financial advisors, and on the UK regulatory side, that the Bank of England governor sent a letter to the G20 warning of an AI-driven financial crisis.

What stands out in this announcement is the order of deployment. Barclays first ran Claude where people review the output again, in an employee search tool and email sorting, built up usage figures there, and is now extending it to the development organization that changes code directly. Once half of a bank's developers are using a coding agent, verification steps such as code review and deployment approval have to be redesigned to keep pace. Whether the 50% target is met by year-end will be confirmed by the next figures Barclays releases at the end of 2026.