As part of VMblog's annual industry-predictions series, Kamiwaza Field CTO and Technology Evangelist James Urquhart lays out his outlook on the next decade of AI, arguing that the defining shift will be from control to collaboration. He notes that 2025 saw organizations move from fascination with agentic AI to real experimentation, exposing a critical gap: our infrastructure for managing intelligence is still designed for static, human-defined systems, while today's AI reasons independently and self-organizes faster than centralized processes can track. Urquhart frames three trends that will chart this transformation through 2030. First, agentic AI will force organizations to re-evaluate access control: role-based models are ill-equipped for a world of unpredictable multi-agent interaction, so security must move from static permissions to behavior-based safeguards, with a 2026 wave of dynamic, relationship- and attribute-based access control, and entirely new frameworks to monitor agent behavior in real time by 2030. Second, he predicts a rise of "hive" architectures and orchestration platforms — a shift from simple agent swarms to structured systems where a central agent coordinates specialized ones, alongside a clean separation between tools for building agents and platforms for running them, much as application development and cloud deployment diverged. These orchestration platforms, he argues, will become foundational infrastructure for AI-driven businesses. Third, development will shift from line-by-line coding toward conversational, outcome-based collaboration with AI agents. Urquhart likens the moment to the server-to-cloud transition: confusion first, coherence later. The themes map directly onto Kamiwaza's platform and its work on ReBAC and orchestration. Read the full article on VMblog.
Source: VMblog — Read the full article