The era of cloud interfaces designed for the eyes and hands of system administrators is ending faster than we could adapt to it. According to Mikhail Shpakov, Head of Development at Timeweb Cloud, user behavior has undergone a tectonic shift: one in three requests to infrastructure documentation is now generated by AI agents rather than humans. Provider logs show that out of 237,000 visits to the documentation section over a two-week period, bots from AI vendors accounted for 32%. Notably, ChatGPT-User accesses help guides 23 times more often than the traditional Googlebot—and it does so not for model training, but to fulfill specific user instructions in real-time.
The entry barrier for automation has collapsed, fundamentally altering the economics of the process. While deploying resources previously required mastery of Terraform or the Command Line Interface (CLI), the combination of Model Context Protocol (MCP) servers and assistants like Claude Code or Cursor now allows infrastructure to be built using plain text. Timeweb Cloud's statistics confirm this: in just two weeks, the agent-specific endpoint processed 112,000 requests, showing a week-over-week growth of nearly 50%. Engineers have stopped studying "how it works" and have begun delegating the action itself to the machine.
Shifting Priorities for Business
For businesses, this represents a radical change in how providers are selected. Agents and humans consume information differently. A human looks for how to link a credit card or configure SSH; a model looks for how to spin up a server or manage DNS. Cloud selection is moving away from marketing and dashboard aesthetics toward API quality and machine-readable documentation. An agent will ignore banners and "pretty charts," but it will scan technical limits and Total Cost of Ownership (TCO) thousands of times before making a recommendation to the user.
Classic UI is becoming a vestigial organ. Providers have spent years investing in dashboard ergonomics only to end up as a "backend" for third-party chatbots.
The New Reality of Infrastructure Management
API methods and text-based instructions are becoming the new cloud storefront. The decision-making subject has shifted from the human engineer to the AI agent. Engineers are being removed from the infrastructure management loop, retaining only the role of the task setter.
We expected neural networks to help engineers work faster; instead, we have reached a point where the engineer is often excluded from the direct management cycle entirely.