Anthropic has announced that three Claude features will be gradually retired by 15 June as the platform moves toward a more capable and reliable VM-based architecture.
The good news is that most migrations will happen automatically. The main change users may notice is that some workflows now depend on code execution and file creation being enabled. Here's a breakdown of what's changing and what it means for Claude users.
1. Artifacts are moving to a VM-based experience
Existing Artifacts will automatically migrate to VM-based Artifacts, so there's nothing users need to manually move or recreate. However, creating and interacting with Artifacts going forward will require code execution and file creation to be enabled.
The upgraded environment supports creating and editing downloadable files such as spreadsheets, presentations, documents, and PDFs directly inside Claude — meaningfully expanding what you can produce in a single conversation.
2. Styles are being replaced with Skills
Custom Styles are also being migrated automatically — this time into Skills. Skills are becoming the new way to customise Claude's behaviour and workflows. Users who previously relied on Styles will continue using similar functionality through Skills, provided code execution and file creation are enabled.
Anthropic also confirmed that the Learning Mode style will migrate into a Skill. For teams and organisations, some workflows can alternatively be handled through custom instructions set at an organisational level.
3. Project file search is changing
Claude is also improving how Projects search and use uploaded files. Under the new architecture, Project file search will depend on the same VM-based execution environment, meaning code execution and file creation must be enabled for the feature to work fully.
According to Anthropic, project files remain available in Claude's context while also becoming accessible through the computing environment — allowing for more advanced workflows across uploaded documents and datasets.
Why Anthropic is making the change
Anthropic says the move is part of a broader shift toward a more capable and reliable architecture built around sandboxed virtual machine execution. The company has continued investing in governance and security controls around this environment, including configurable network access and restricted domain access.
Anthropic's engineering team has also shared details on how Claude's sandboxing works, including filesystem isolation and network isolation designed to reduce risk during code execution. In short: the platform is being re-platformed onto something that can do much more, more safely.
What users should do before 15 June
Most users won't need to manually migrate anything, but it's worth checking whether code execution and file creation are enabled if you regularly use:
- Artifacts
- Custom Styles
- Project file search
- Advanced document workflows
Without those capabilities enabled, some migrated features may no longer function as expected after the retirement deadline. For organisations on Team or Enterprise plans, this is also a good moment to review which capabilities are turned on at the workspace level — settings can vary by admin policy.
Final thoughts
This update signals a bigger shift in how Claude is evolving. Rather than treating features like Artifacts and file search as standalone tools, Anthropic is consolidating them into a unified VM-powered workflow experience. For users, that likely means more capable automation, richer document handling, and more advanced AI-assisted workflows going forward.
If you'd like help auditing your team's Claude setup before the June 15 deadline — or designing workflows that take advantage of the new VM-based capabilities — the Okiru team is here to help. Get in touch via okiru.co.za.