Anthropic launches an engineer residency for enterprise AI deployments

Anthropic wants to train 10,000 engineers who can put AI to work inside companies by the end of 2027. On October 2, it announced Claude Frontier Academy with a $100 million commitment and a training model that extends beyond the classroom: participants return to their employers to lead a real project.

The academy begins with a residency for Frontier Deployed Engineers, whose work combines software engineering with a deep understanding of the business. It involves much more than choosing a model and writing good instructions. An application must connect to existing systems, access the right data, pass security review and serve the people who will use it every day.

That is the promising part of the initiative. Learning takes place close to the problems that tend to emerge after a successful demo, when a team has to turn an idea into something it can keep running.

Four days in person, followed by 12 weeks at work

The in-person intensive takes place in New York, San Francisco and London. Engineers spend three days building a system for a simulated enterprise, from the initial request through security review and operational handover. On the fourth day, a practical assessment introduces a new scenario. Successful participants earn the Claude Resident Engineer credential.

The next stage lasts 12 weeks. Engineers lead a Claude deployment at their own organization with mentoring and support from Anthropic practitioners. A final assessment leads to the Claude Frontier Deployed Engineer credential, with the first awards expected in early 2027.

Employers nominate participants and should contact their Anthropic account team or Partner Account Manager to check eligibility. Candidates need strong engineering fundamentals, LLM experience and a record of helping colleagues adopt AI. Prior experience building agents is not required. Travel and accommodation arrangements still need to be clarified with the program organizers.

A nomination also calls for an internal conversation. Participants arrive with a defined project, and their employers need to make room for them to deliver it. Protected time, system access and involvement from the business team can determine whether the residency produces a useful application or another presentation.

The hard part is making the pilot part of everyday work

Companies have expanded access to AI, but delivering consistent value remains harder. In Accenture's Pulse of Change, published in July 2026, only 23% of executives reported widespread, sustained business value from AI. The surveys covered 3,000 leaders and 3,000 employees between April and June.

Insufficient skills also emerged as the biggest barrier to AI integration in Deloitte's 2026 State of AI in the Enterprise research. About one in five organizations had mature governance for autonomous agents. The survey covered 3,235 leaders in 24 countries in August and September 2025.

Those findings help explain the appeal of training tied to work. A course teaches techniques; a deployment requires deciding where to apply them and handling the consequences. In customer service, for example, generating a helpful answer is only part of the job. The team must decide when to hand a case to a person, what to record and how to recover from an error without harming the customer.

Learning to build the application includes learning what to do when it fails.

The training has to leave capability behind

The residency brings learning closer to delivery, but its impact still needs to show up in projects. A system can work during training and encounter difficulties once it serves real users, data and exceptions. The employer's evaluation needs to continue after the credential.

There is a strategic choice for participating companies too: benefit from working alongside Anthropic engineers while developing practices that remain useful as the technology changes. Evaluating outputs, controlling permissions and preparing operations remain responsibilities even when a team changes models.

The benefit could extend beyond the first application. An engineer who teaches colleagues, documents decisions and leaves a process others can maintain helps the company build its own capability. That work needs to continue after the 12 weeks end.

References: Claude Frontier Academy launch, program structure and participation, Accenture Pulse of Change and Deloitte State of AI in the Enterprise 2026.