What if AI cleaned up a bank's 'complex legacy'? Barclays' unique experiment

Imagine this: decades ago, complex and outdated computer programs were built to handle banking tasks. Today, bank employees struggle daily with these ancient systems to process data. But what if a smart AI assistant appeared, capable of instantly understanding and neatly organizing these complex bundles of code? Recently, the global financial institution Barclays has been making this transformation a reality.

Why does this matter?

In the financial sector, “digital transformation” is not just a buzzword. Core banking systems are built on layers of complex code that have accumulated over decades. Modernizing these systems is as difficult and costly as changing the wheels of a moving train.

Barclays’ enterprise-wide adoption of Anthropic’s AI model, “Claude,” is a highly strategic decision. The plan is to efficiently upgrade outdated systems and drastically reduce unnecessary operational costs by leveraging the power of AI (Source: Barclays scales Claude to upgrade operations). This illustrates how the competitiveness of modern banks depends on how quickly they can adapt to the digital environment.

Simply put, what kind of technology is it?

You can think of “Claude” as an “AI assistant”—a highly skilled software engineer and a meticulous analyst.

When we remodel a house, it’s hard to tell where the problems are just by looking at the blueprints, but a veteran architect can instantly spot structural flaws. Claude is the same. In particular, the latest model, “Claude Opus 4.6,” can grasp vast amounts of information—up to “one million tokens”—at once (Source: Introducing Claude Opus 4.6). A “token” here refers to the unit of information volume that the AI can read and understand at one time. This means it is capable of reading through giant bundles of banking code consisting of tens of thousands of lines at a glance, judging for itself what is outdated and where errors have occurred.

How far has it progressed?

Through this strategic partnership, Barclays is deploying Claude across its global operations (Source: Barclays expands use of Anthropic’s Claude AI). The most visible change is in the development environment. Barclays has set a goal for 50% of all its developers to be using ‘Claude Code’ by the end of 2026 (Source: Barclays scales Claude to upgrade operations). ‘Claude Code’ is an AI tool that acts like a partner, assisting developers as they perform complex coding tasks.

Claude has already recorded explosive user growth globally, establishing itself as a powerful collaborator for practitioners in finance and various other industries (Source: Q1 2026 AI Intelligence Report).

AI’s Take: Future Outlook

For traditional financial institutions, innovating legacy systems through AI is no longer a choice but a matter of survival. When high-performance AI like Claude becomes a partner to practitioners, banking productivity will skyrocket.

If Barclays’ experiment succeeds, other financial institutions are likely to follow suit. We will see increased use of “agents” that go beyond simple conversational chatbots—tools that can perfectly understand system internals and fix code autonomously. If the mobile banking app you use becomes faster and smarter in the future, it might just be because AI is working hard in the background, cleaning up the legacy systems.

References

1. Barclays scales Claude to upgrade operations and improve client experience
2. Barclays expands use of Anthropic’s Claude AI across global operations
3. Claude on X: “Introducing Claude Opus 4.6. Our smartest model got an upgrade.”
4. Comscore’s Q1 2026 AI Intelligence Report Shows Scale of Claude’s breakout growth