Barclays is expanding its collaboration with Anthropic to integrate Claude across the bank's operations, according to an announcement published on Anthropic's website. Barclays says it expects Claude Code adoption to reach 50% of its developer population by the end of 2026, rising to a majority of software engineers in 2027, with the stated aims of accelerating software development, modernizing legacy systems and improving operational efficiency. The announcement cites two existing deployments. A Colleague Knowledge Assistant, live since 2025 and built on a retrieval-augmented generation architecture, helps Barclays UK employees find information when supporting the bank's stated 20 million-plus UK retail customers; Anthropic says more than 16,000 colleagues have adopted it and it has handled over one million searches. Within Barclays' Global Markets business, Claude models classify, enrich and route incoming emails, processing approximately 120,000 emails each day. Group Co-Chief Operating Officers Anne Marie Darling and Craig Bright are quoted describing the work as reimagining processes and embedding AI within how technology is built, tested and secured. The announcement states the bank applies governance, security controls and human oversight to AI use cases. The figures and claims originate from Anthropic, a vendor announcing its own partnership; independent verification of the outcomes was not provided.
- Barclays expects Claude Code adoption to reach 50% of developers by end of 2026 and a majority of engineers in 2027
- A Claude-powered Colleague Knowledge Assistant, live since 2025, has been adopted by 16,000+ colleagues and handled over one million searches
- Claude models process approximately 120,000 emails per day in Barclays' Global Markets business
- The bank states it applies governance, security controls and human oversight to AI use cases
What it means for you
A very large, heavily regulated bank is rolling AI into real operational work — triaging emails, helping staff find answers, and writing code. The headline numbers come from the AI vendor's own announcement, so treat them as claims, not independently audited results. For most organizations the useful lesson isn't the tool choice; it's what Barclays says it did around the tool — governance, security controls, and keeping a person on the decisions that matter.
Try this
Pick one high-volume, repetitive intake task in your shop — inbound email, support requests, document sorting — and spend an hour mapping how it flows today: who touches it, where it stalls, which steps are judgment and which are rote. That map tells you whether AI could help or whether the process itself is the problem.
Who should care
Operations leaders at organizations with high-volume request handling (email triage, internal help desks) and anyone managing a software development team weighing AI coding assistants.
Skip this if
You're a small team without a large, repetitive intake queue or an in-house engineering group — a billion-dollar bank's rollout plan doesn't change what you should do this quarter.
Sources: Anthropic — read the original