AI agents are slipping into roles traditionally held by humans, such as updating hospital records and inspecting factory lines, but there’s a growing issue: enterprise identity and access management (IAM) systems weren’t built for them. While AI’s potential is vast, it’s stuck in pilot phases largely due to trust and governance issues, not technological capability. As Cisco President Jeetu Patel noted at the RSAC 2026, a mere 5% of enterprises have moved beyond the pilot stage despite 85% dabbling in AI agents. The root of this stagnation? A significant gap in secure identity governance that can’t keep up with non-human entities at machine speed.
### What AI Agents Can Actually Do
In healthcare, AI agents assist doctors by updating electronic health records and suggesting prescriptions in real time. Similarly, in manufacturing, AI-driven computer vision systems conduct quality control faster than any human could manage. These agents generate their own identities, which traditional IAM systems struggle to manage effectively. Unlike human users, AI agents require IAM systems to dynamically adjust permissions and access controls at speeds that match their operational tempo. The challenge isn’t just about having the right tools but rethinking the entire architecture of identity management to accommodate these autonomous entities.
### The Competitive Context: IAM’s Shortcomings
Despite the demonstrated capabilities of AI agents, the enterprise sector remains cautious. According to IANS Research, the current state of role-based access control is insufficient even for human identities, let alone AI. IBM’s 2026 X-Force Threat Intelligence Index reported a 44% rise in attacks exploiting public-facing applications, highlighting the vulnerabilities in existing IAM frameworks. The lack of mature identity governance is a bottleneck that prevents AI from entering mainstream production environments. The competitive edge is being sharpened not by who has the most advanced AI capabilities but by who can effectively govern these digital entities without compromising security.
### Real Implications for Founders, Engineers, and the Industry
For founders and engineers, the critical takeaway is that deploying AI agents requires a foundational shift in how identity and access are managed. Michael Dickman, SVP and GM of Cisco’s Campus Networking business, emphasizes the need for a trust framework that can handle the complexity of AI agents. The network’s ability to see actual system-to-system communications is crucial for cross-domain correlation and enforcing policy at machine speed. Engineers must design systems that prioritize trust and security from the outset, rather than as an afterthought. This means developing IAM solutions that can dynamically adapt to the fluid nature of AI agents.
### What Happens Next?
As enterprises grapple with integrating AI agents into their operations, the focus will need to shift towards developing IAM systems that can handle the unique challenges posed by these non-human entities. For those in the tech industry, this means there’s a growing opportunity to innovate in the realm of identity governance. Founders and engineers who can create solutions that bridge the trust gap will find a receptive market. Investors should look for startups and companies that are not just dabbling in AI but are addressing the structural identity issues that keep AI from reaching its full potential in enterprise environments.
