CNBC reporters with no coding experience built a functional clone of Monday.com's core features in under an hour using Anthropic's Claude Code. The vibe-coding exercise cost $5-15 in compute credits and produced a working project management dashboard with team assignments, status tracking, calendar integration, and email connectivity.
The test comes as software stocks post their worst six-month relative performance against the S&P 500 on record. This week's selloff saw EPAM Systems drop double digits, while Gartner and Accenture fell over 20%. The trigger: Anthropic's release of nearly a dozen Claude Cowork plugins for sales, legal, and data analysis.
Silicon Valley sources tell CNBC the most exposed companies are those that "sit on top of the work" - tools like Atlassian, Adobe, HubSpot, Zendesk, and Smartsheet that aren't core infrastructure. These productivity layers can be replicated by AI agents following plain-English instructions.
Systems of record appear more defensible. Salesforce anchors enterprise data architectures that can't be cloned in a weekend project. Cybersecurity stocks like CrowdStrike and Palo Alto benefit from network effects no customer wants to rebuild.
William Blair analyst Andrew Nicholas calls the broader selloff "particularly irrational" for data firms like S&P Global and Moody's, which have viable AI business models versus pure professional services plays.
The real question: Does fast prototyping equal enterprise replacement? The CNBC clone worked for personal project management. Scaling it to handle enterprise permissions, compliance, integrations, and support is different work entirely. That gap may separate genuine disruption risk from market panic.
Notably, the compute costs could drop further as data center capacity expands. For now, the test demonstrates that certain software categories face legitimate pressure from AI coding tools, while others benefit from structural moats the market is currently ignoring.