Anthropic and DXC bring Claude into the regulated systems banks and airlines run on
A global alliance puts Claude agents to work modernising insurance, banking and security operations.
Inteeka · 11 June 2026 · 4 min read
Most AI announcements are about the next model or the next benchmark. This one is about plumbing, and that makes it more interesting, not less. Anthropic and DXC Technology have announced a multi-year global alliance to embed Claude inside the kinds of systems that quietly run modern life: the ones banks, airlines, insurers, manufacturers and government agencies depend on. These are not greenfield apps. They are the regulated, compliance-heavy, decades-old systems that real organisations are nervous about touching. The news is that someone has decided to touch them, carefully and at scale.
What was announced
DXC is a large global IT services company (around 115,000 employees across 70 countries) and it already runs Claude across its own operations. The alliance turns that internal use into a service it offers clients, with four declared focus areas: insurance, where it will build agentic solutions and modernise core systems; Modernization as a Service, which analyses and refactors legacy codebases; cybersecurity, framed as an “always-on security engineer” subagent for security operations centres; and application services, where Claude agents are embedded into the work of maintaining existing applications.
To deliver this, DXC says it will train tens of thousands of Claude-certified forward-deployed engineers, with certification running through Anthropic Academy on a DXC-specific curriculum, and it has joined the Claude Partner Network. The headline proof point comes from DXC’s own platform, OASIS, which launched in April 2026 and now serves more than fifty customers: on that platform, Claude generated over 95% of the code and accelerated software development by roughly tenfold.
Why it matters for businesses
The significant detail is the choice of battleground. For two years the easy wins for AI have been on the edges: drafting copy, summarising documents, answering support questions. This alliance aims at the opposite: mission-critical, regulated systems where a mistake is expensive and an auditor will eventually ask questions. That is a deliberate signal that the technology, and the practices around it, are considered ready for the parts of a business that cannot simply be rebuilt over a weekend.
Two threads are worth drawing out:
- Legacy is now in scope. The Modernization-as-a-Service framing treats old codebases as something to be read, understood and refactored by agents rather than quietly tolerated. The old assumption, that legacy systems are too risky to disturb, is being challenged directly.
- Compliance is the design constraint, not an afterthought. The announcement leans on strict security and compliance requirements as the thing being solved for, not a hurdle to apologise about. That reframes AI in regulated settings from “can we?” to “how do we do this responsibly?”
What to do about it
Not every organisation needs a tens-of-thousands-strong engineering programme, and the eye-catching figures are specific to DXC’s own platform rather than a promise about anyone else’s. The useful takeaway is the playbook underneath, which scales down cleanly. Pick a single legacy or compliance-heavy process (claims handling, a reconciliation step, a slice of a security workflow) and treat it as a modernisation project with a measurable outcome, rather than a vague ambition to “add AI”.
Then put the guardrails in before you scale, not after the first incident: clear scope, validated inputs, audit trails, and human approval where the stakes are high. The lesson of this alliance is that the careful path and the ambitious path are the same path. You earn the right to automate more by doing the first thing well and being able to show your working.
The takeaway
When the systems banks and airlines run on are considered fair game for AI agents, the question for most businesses shifts. It is no longer whether AI belongs near the serious parts of the operation, but how to bring it there without breaking anything that matters. The reassuring part is that this is known work: scope tightly, modernise one thing at a time, build the compliance in from the start, and measure as you go.