What identity means in the age of agentic AI: Insights from Think 2026

This week atThink 2026, IT and business leaders gathered to discuss identity in the age of agentic AI, and one theme came through clearly.Agentic AIis reshaping how enterprises operate, but traditionalIAMwas never built for the agents andnon-human identities (NHIs)now driving that change.

Whether you joined us in Boston or are catching up from afar, here are the biggest takeaways for identity and security leaders coming out of Think.Agentic AI is a fundamentally new kind of system.Unlike traditional AI that waits for prompts, agents act on their own. They make decisions, execute tasks and call tools across systems with minimal human oversight. A single user request can fan out into dozens of agent actions in seconds.The potential upside is enormous.According to the IBM Institute for Business Value, 53% of executives expect AI to transform business models in their industry by 2030. Organizations leaning into AI-first are projecting 70% productivity gains, 74% faster cycle times and 67% faster project delivery.But traditional IAM systems weren’t built for this—in fact, 92% of organizations aren’t confident their legacy IAM tools can manage the risks that come with NHIs and AI agents1. These systems were designed for humans who log in, hold sessions and get reviewed quarterly, not for agents that act autonomously, continuously and at machine speed.At the same time, non-human identities are exploding.There are now 45 to 90 NHIs for every human identity in enterprise environments2, and the ratio grows with every new agent deployment. Most organizations have no real visibility into which agents are running, who deployed them or what they’re authorized to do.It’s an organizational problem almost as much as a technological one.NHIs are largely created and managed by development teams, while security teams and CISOs still focus on humans. In a recent IBM survey of 1,100 industry professionals, nearly three-quarters said their platform and development teams don’t work well with their security teams. This means the explosion of agentic AI is happening across an internal divide that hasn’t been bridged.IBM’s response: control at every action.Just this week, IBM unveiledVault 2.0, which modernizes identity-based security at scale for the agentic era, and a unified Verify and Vault solution, IBM’s end-to-end answer for securing both human and NHIs across the enterprise.