AI Agents Transform Cybersecurity and Support

A New Era for Cybersecurity with AI Agents

Cybersecurity companies are increasingly turning to artificial intelligence agents to lighten the workload of their analysts. These agents allow experts to focus on more complex attacks by automating the sorting of alerts. For example, Huntress, a company specializing in threat detection, has managed to reduce investigation time and analyst workload by 90%. This technological advancement enables more effective management of cyber threats.

However, AI agents are not without limitations. They often require human supervision, especially in high-risk situations, as they can struggle with complex tasks. Security companies, faced with a constant increase in threats and a shortage of qualified personnel, are adopting these AI agents to maintain their effectiveness. Unlike generative AI tools like ChatGPT, AI agents are designed to perform specific tasks and follow multi-step workflows.

Growing Adoption of AI Agents

The shift to automated workflows is already well underway. According to a McKinsey survey conducted in 2025, 62% of the companies surveyed reported experimenting with AI agents. In the cybersecurity sector, this trend is also on the rise. A study conducted by ISC2, a nonprofit organization specializing in cybersecurity, revealed that 30% of professionals are integrating AI security tools into their operations. These systems are evolving into tools similar to agents capable of managing processes once reserved for human analysts.

Huntress and Threat Detection

Huntress has implemented nearly 20 AI agents in its security operations center, which processes alerts for 240,000 clients. Eric Stride, Huntress's security director, explains that these agents automate investigations that were previously conducted manually by a team of 50 people. When a suspicious signal is detected, an orchestration agent, acting as an AI supervisor, activates 12 sub-agents to extract and analyze data and identify evasion techniques.

This system allows for determining whether an activity is malicious or benign, with ambiguous cases being forwarded to a human analyst for verification. The process, which used to take between 20 and 30 minutes manually, can now be completed in just a few minutes. Stride emphasizes that this system has reduced the workload of analysts by 90% for more than a third of investigations, generating about 10,000 incident reports per month. Analysts can thus focus on more complex attacks, benefiting from what he calls their "Iron Man suit."

AI Agents in Customer Support

At DNSFilter, Mikey Pruitt, head of AI labs, has introduced an AI agent into the customer support team, which consists of fewer than 10 engineers. This agent now handles all level 1 tickets. When a customer submits a ticket, the AI agent categorizes it according to its complexity and resolves common issues based on internal documentation. More complex tickets are forwarded to human engineers. This process takes about four minutes, and an AI agent can handle 60 tickets per week, compared to 35 for a human, freeing up to three hours per week for engineers.

The Challenges of AI Agents

Despite their advantages, AI agents have limitations. At Huntress, Stride notes that agents can fail on vague tasks, sometimes producing inconsistent responses. They are effective for repetitive tasks but less so for complex threats like ransomware. Additionally, they cannot make high-risk decisions without human intervention. At DNSFilter, the AI agent, limited to internal documentation, has occasionally provided incorrect advice, such as bypassing a reseller partner, which required adjustments.

A Compelling Economic Strategy

Despite these challenges, the cost savings achieved through AI agents are significant. Pruitt indicates that an AI agent costs between $15,000 and $16,000 per year but performs the work of two full-time support engineers, resulting in savings of $200,000 annually. This strategy allows DNSFilter to reduce the hiring of entry-level staff. As AI agents become more sophisticated, they could take on more complex support tasks, transforming the customer support team into roles for engineers or quality assurance specialists. For now, companies are using AI agents to boost productivity without increasing headcount, aiming to make a team of 150 operate as if it were 500 by the end of the year.