Why Agentic-First Startups Won't Disrupt Enterprises as Fast as Everyone Thinks | Kris Lovejoy

What It Covers

Kris Lovejoy, global strategy leader at Kyndryl (IBM's IT infrastructure spinoff), explains why agentic AI adoption in enterprises will take until roughly 2031 to reach scale, citing infrastructure gaps, security risks, compliance demands, and the absence of horizontal workflow integration across business functions.

Key Questions Answered

• •Agentic AI Adoption Timeline: Lovejoy predicts that by 2031, approximately half of traditional IT systems administration tasks — specifically line one and line two support functions — will be handled by agentic AI, with humans either in the loop or overseeing operations. Enterprises should plan infrastructure modernization now to meet that five-year runway.
• •IT Service Management as Entry Point: Of 34 ITIL-defined IT processes, roughly 20 are candidates for agentic automation, covering areas like patch management, incident resolution, and configuration management. Kyndryl's model shows this can reduce IT service management costs by up to 90%, creating a "modernization dividend" to fund broader infrastructure upgrades.
• •Startup Disruption Is Slower Than Advertised: Agentic-first startups face a hard ceiling when selling into enterprises because demos rarely survive contact with SOC 2 compliance requirements, legacy SAP integrations, and multi-cloud environments. Consumer-facing B2C applications will see faster agentic disruption than B2B enterprise software, where data integration complexity blocks entry.
• •Context Gap Is the Core Security Risk: The most common agentic AI failure mode is not sophisticated cyberattacks but misconfiguration — an agent patching a protocol that was intentionally left unpatched due to a 40-year-old legacy dependency. Enterprises must build knowledge bases capturing *why* systems are configured as they are, not just *what* the configuration is.
• •Workforce Shift Requires Apprenticeship Models: As agentic AI eliminates entry-level SOC roles, organizations face a pipeline problem — fewer junior analysts means fewer future senior engineers. Lovejoy recommends co-funded apprenticeship programs between universities, businesses, and public sector bodies that put students on live systems from day one, producing job-ready level-two engineers at graduation.