CommBank engineers deepen expertise through Anthropic's Claude Frontier Academy

The first stop is a four-day intensive built around simulated enterprise deployments: taking an AI system from customer requirements and solution design through testing, security review, compliance and production. After that, the initial group of 20 CommBank engineers moves into a 12-week residency leading real Claude work inside the bank, with more engineers joining as the programme expands.

What happened

Anthropic launched Claude Frontier Academy on 2 October with a $100 million commitment and a goal of training 10,000 Frontier Deployed Engineers by the end of 2027. Commonwealth Bank of Australia is in the first cohorts alongside Accenture, Bain, Capgemini, Deloitte, McKinsey, Morgan Stanley and Novo Nordisk. Cohorts run in San Francisco, New York and London, and places are by nomination, not open enrolment.

Residents earn a Claude Resident Engineer credential after the intensive. They are assessed again on a final practical before they earn the Claude Frontier Deployed Engineer badge, which Anthropic expects to award for the first time in early 2027.

Rodrigo Castillo, CommBank's Group Chief Technology Officer, said the bank's engineers already use AI tools every day and that its teams "have produced up to 3x more code changes in the past year." The residency, he said, "takes them deeper, with hands-on experience building, securing and shipping alongside peers at the frontier."

Why the format matters

Most enterprise AI training stops at a licence and a prompt course. You've probably seen the dashboard that says 98% complete. Completions are cheap. What CommBank has signed up for is a cohort, a residency and an assessment, which looks a lot more like how a hospital trains a registrar than how a bank rolls out software.

Anthropic says as much. It describes the residency on the medical model: learn from practitioners, practise on realistic cases, and get assessed on your work before you practise alone. The assessment is the part most corporate academies skip.

The cohort size is worth noticing too. One engineer sent away on a course comes back with new skills and no one to use them with. Twenty engineers who trained together come back with a shared vocabulary, a peer group to review each other's work, and the start of an internal practice that can teach the next wave.

What this means for L&D

Traditional training separates learning from application. People attend a workshop, return to their desk and hit the gap between theory and the messy reality of production systems. A residency closes that gap by making the live project the classroom.

If you run AI upskilling, three questions are worth asking. Does the programme include supervised time on real work, not just content? Is there an assessment that someone could actually fail? And are you building cohorts, or sending individuals who come back alone?

Most organisations cannot send engineers to an external residency. The principle still travels. Pick a live AI project, staff it with a small cohort, pair them with someone who has shipped before, and judge them on what goes into production. That is how capability spreads beyond the people who happened to be in the room.

Sources:

• CommBank, 3 October 2026
• Anthropic, 2 October 2026