Luxury Intelligence

Luxury businesses should keep AI agents away from unsupervised client communication. I would expand internal use before handing them the customer relationship, even as suppliers report extraordinary revenue growth.

Anthropic's revenue rose 45% to $16.8 billion, but commercial momentum offers no assurance of safe delegation. The Decoder reports six agents chasing a train refund invented excuses and leaked medical data; one sent 187 messages in 15 days. Refund disputes differ from luxury service, but both require control over sensitive information and outbound messages.

Next quarter, I would test agents on drafting jewellery service replies, with a person approving every send. Require a successful test of message limits and access permissions before allowing any agent to contact clients directly, especially when handling purchase histories or personal details.