AI Agents Reshape Security Operations: SOCs Pivot Toward Human-Machine Collaboration

The landscape of modern cybersecurity is undergoing a profound transformation as artificial intelligence transitions from a conceptual buzzword to an operational necessity. According to the latest findings from the security startup Strike48, the vast majority of security leaders—a staggering 87%—are currently in the process of redesigning their security operations centers (SOCs) to integrate AI agents. While this shift has sparked widespread anxiety regarding the potential displacement of human talent, the reality revealed by the report, The State of Agentic Security H2 2026, suggests a more nuanced evolution: roles are not being eliminated, but rather significantly adapted to meet the demands of an automated future.

The report, which draws on data from 534 security and IT leaders across the United Kingdom, Canada, and the United States, provides a clear window into the rapid adoption of agentic technologies. The surveyed organizations, all of which employ at least 200 staff members and maintain a dedicated security operations function, indicate that AI agents have firmly entered the mainstream. An impressive 74% of respondents confirmed they have already deployed agentic technology within their SOC environments. This represents a dramatic acceleration in adoption, rising sharply from 36% in May of the same year. Furthermore, the integration is deepening; nearly half (49%) of these organizations are now utilizing AI agents across multiple operational use cases, signaling that the technology is moving beyond pilot programs and into the core of daily security workflows.

The Evolution of the SOC Analyst Role

For years, the cybersecurity industry has grappled with the fear that AI would eventually render entry-level SOC analysts obsolete, effectively closing the door on a vital career pipeline for junior professionals. However, the data from Strike48 offers a more optimistic outlook. Rather than facilitating mass layoffs, organizations are choosing to redeploy their Tier-1 (L1) analysts into more sophisticated functions that leverage their existing security knowledge while augmenting their capabilities with AI.

The transition strategy is multi-faceted. Half (49%) of the security leaders surveyed reported that they are actively hiring for new, specialized roles that did not exist or were not prioritized just a few years ago. Positions such as AI detection engineering and automation specialists are becoming increasingly critical as SOCs look to manage the complex interplay between human oversight and autonomous systems. Simultaneously, 42% of respondents are focusing on internal mobility, retraining L1 analysts to serve as supervisors for these AI agents. This transition emphasizes the "human-in-the-loop" model, where the analyst evolves from a frontline alert processor into a high-level manager of autonomous security assets.

Other organizations are pivoting their talent strategy to maximize the value of their human capital. About 38% of teams are moving L1 analysts into dedicated investigation roles, allowing them to focus on the complex, human-centric nuances of threat hunting that AI may still struggle to fully interpret. Additionally, 30% of organizations are even expanding their teams by bringing on more entry-level analysts. This suggests that for many firms, the introduction of AI is not a signal to shrink the workforce, but rather an opportunity to raise the operational baseline and handle a greater volume of security alerts that previously went unaddressed due to capacity constraints.

This trend is a necessary counter-narrative to earlier industry concerns. In September, a report from Swimlane highlighted that 24% of SOC analysts felt that emerging technologies had hindered their professional development by limiting their ability to build foundational skills. Furthermore, one in 10 respondents expressed fears that junior analysts would find fewer opportunities to gain the hands-on experience required to advance their careers. The Strike48 data suggests that while these concerns are valid in terms of the changing nature of the work, the industry is consciously moving toward a model where human oversight remains a cornerstone of the SOC, effectively preserving the career path for those willing to adapt.

Most SecOps Leaders Are Changing the L1 Role as Agent Use Grows

Building Trust for Autonomous Security

While the deployment of AI agents is accelerating, the report uncovers a significant hurdle that is currently constraining the potential of these technologies: a lack of organizational trust. Despite the high rates of adoption, the vast majority of security teams remain hesitant to grant their AI agents full operational freedom. Only 30% of organizations report that they allow their agents to act with high or even full autonomy.

The majority of deployments remain under strict human control. Nearly half (47%) of respondents state that they keep humans in the loop, requiring them to execute actions or at least provide pre-approval before an agent can take any significant step. A further 23% of organizations allow their agents to act, but only within very narrowly defined, rigid boundaries. This conservative approach to automation is directly linked to the maturity of the deployment process and the level of governance established within the SOC.

Tim Leehealy, co-founder and president of Strike48, suggests that the current skepticism surrounding AI agents is less a failure of the technology itself and more a failure of implementation strategy. "Most teams start the same way: one agent, one long prompt telling it to act like a Level 1 analyst, and a pile of tools bolted on," Leehealy explained. "Then it meets a real alert queue, and you find out a single prompt can’t carry an entire triage process. In my view, this is why distrust of agent outputs is highest for teams with just one use case."

According to Leehealy, the path toward greater trust and higher autonomy lies in breaking down complex security processes into manageable, granular tasks. When organizations attempt to treat an AI agent as a "black box" that can handle an entire end-to-end triage process without proper structural design, the results are often unpredictable, leading to the distrust observed in the survey data.

"I’d argue the models are mature enough," Leehealy noted. "What’s immature is how most teams use them. The SOC teams that get to agent autonomy break the work into narrow jobs and fixed, deterministic workflow steps."

As organizations continue to refine these workflows, the relationship between human analysts and AI agents is expected to become more symbiotic. The industry appears to be moving toward a future where security operations are defined by this partnership. In this environment, the most successful SOCs will not be those that attempt to replace human intuition with automation, but those that empower their analysts with the tools to direct, supervise, and improve autonomous agents. As these processes become more standardized, trust in AI outputs will likely rise, enabling security leaders to push the boundaries of what is possible in the ongoing battle against increasingly sophisticated cyber threats. For the workforce, this means the future of the SOC is not one of obsolescence, but one of transformation—where the ability to orchestrate, supervise, and adapt to AI-driven systems becomes the most valuable skill a security professional can possess.