‘Our identity agent represents 1,000 brains of someone who knows identity’ - Performanta CEO talks security agents

The MSSP is going to market with four security agents, working with Microsoft – and it is planning further rollouts

As solution providers increasingly experiment with agentic AI options from vendors, some are deepening the collaboration and DIYing their agents.

Case in point: MSSP Performanta, which has deployed two proprietary security agents in collaboration with Microsoft and is in the process of deploying two more.

The first of these, an email security agent, is designed to hunt, triage and take action against phishing attempts, social engineering, malware and other email-based threats. According to Performanta CEO Guy Golan, the agent can dramatically reduce manual workloads for security teams.

“For every 10,000-user organisation, you typically need three or four people to handle email security on a daytime basis,” Golan explains. “If you’re a 24/7 multinational, you’d need three shifts of that – easily 12 people. Our agent doesn’t take that down to zero, but it does take it from 12 to two.”

The second agent – Performanta’s identity security agent – tackles what Golan calls the “new perimeter” in cybersecurity.

“The reduction of attack surface is through understanding the human element,” he says.

“But not many SOC analysts understand identity. Our identity agent essentially represents 1,000 brains of someone who really knows identity, acting as a companion to other analysts and flagging any identity-related violations – lateral movement, identity theft, leavers, impersonations, you name it.”

These two agents are already live, and Performanta is currently selling them via its own channel and through Microsoft’s marketplace and sellers. Golan says Performanta has signed up nine MSPs so far, with strong interest from those looking to increase their per-seat revenue and offer AI-enabled security as a managed service.

“The go-to-market is through Microsoft and through our channel,” Golan says.

What sets Performanta’s agents apart, Golan argues, is architecture. Unlike many agents, which are often limited to a narrow, siloed function, Performanta’s agents talk to each other, forming what he described as “an emulated environment as opposed to isolated tasks.”

“We don’t think about agents as functions. We think of them as clusters of humans, or even autonomous systems,” Golan says.

“Our architecture allows agents to plug in and out, communicate, and build a proper decision tree with processes and dialogue that lead to a definitive answer to keep clients safe.”

Performanta is also building a threat-hunting agent that Golan described as “a revolution on its own,” using agent collaboration to proactively seek out threats and emulate the detective work of human analysts.

Additionally, the company is working on an agent monitoring service, AI Guardian, which will focus on ensuring agent behaviour remains safe and controlled.

Golan foresees this becoming increasingly critical as more organisations adopt agentic AI.

“We foresee in the next six to 12 months that companies will need to monitor not just user behaviour, but agent behaviour,” he says.

“We believe agent behaviour analytics will become just as important.”

Looking further ahead, Performanta sees potential for these agents to expand beyond Microsoft’s ecosystem, with Golan confirming they own the intellectual property outright.

“Nothing stops us from launching these agents on Google, Amazon, Facebook platforms,” he says.