Deal Adds Natural Language, Multi-Agent RAG Tech to Autonomous Security Platform
Torq bought an Israeli retrieval-augmented generation startup to enable independent agents to prioritize, investigation and resolve threats in real-time with minimal human oversight.
The New York-based autonomous security operations vendor said the purchase of Revrod will enable autonomous agents to handle security tasks from initial triage to resolution independently and intelligently, said co-founder and CTO Ofer Smadari. Customers can expect smarter and faster threat mitigation with the added promise to building custom “hyper-agents” using natural language prompts.
“They looked for my advice about the market, and I said, ‘Hey, selling technology is not easy, you need to build a lot more,'” Smadari told Information Security Media Group. “And then we started thinking about, ‘Hey, why don’t you join us? You can help us build our new HyperSOC platform.'”
Revrod, founded in 2023, employs roughly 10 people and has remained in stealth mode since its inception. The team under Torq will be led by Noam Cohen, who previously worked at Explorium, Tetrix Health and Technion before joining Revrod in April 2023 as chief product and technology officer.
Smadari saw the Revrod team as not only capable but as already having a head start with a $6 million seed round and a working product. Building similar capabilities in house would have taken much longer, and recruiting AI engineers of their caliber in today’s market was no easy feat. Smadari said the deal will help Torq leapfrog competitors by acquiring proven AI talent and production-grade infrastructure.
What Makes Revrod Different From Traditional Automation
Unlike traditional automation, which often relies on static playbooks or hard-coded workflows, Revrod’s new RAG-powered engine brings human-like, flexible intelligence into security operations, he said. Multi-agent systems are capable of performing level-one investigations, triaging alerts, prioritizing incidents, and selecting the right integrations and tools for mitigation without human intervention, he said.
“They know how to prioritize things, they know how to do a level-one investigation and triage by themselves,” Smadari said. “And more importantly, they know how to choose the right tool to operate and solve the right need.”
The deal will change a typical security workflow from dozen of steps requiring human input at various stages to a few AI-driven actions completed autonomously with zero human touch, Smadari said. Clients are already using the new AI infrastructure in production environments, where the system can ingest an alert, gather context, evaluate the threat and resolve the issue all without analyst intervention, he said.
“Things will be smarter, faster and help our customers to deal in the battlefield with new attacks and weekly use cases with fewer resources,” Smadari said. “We have technology that can work for our customer only from server – all day – completely online.”
The Next Frontier in Autonomous SOC Technology
Torq plans to debut customizable, intelligent agents that customers can create themselves using plain English instructions, which will enable security professionals to describe use cases such as phishing detection and malware remediation and get the platform to automatically build and deploy an agent to address it. This represents the next frontier in autonomous SOC technology: an AI system that builds itself around your needs.
“You don’t need to deal with building another workflow, so no code or anything else in the future,” Smadari said. “So, you can actually build your use cases and build your own agents inside the top level in that.”
Torq is actively scouting for other young, AI-focused teams that align with the company’s mission of transforming security operations, according to Smadari. Ideally, he said Torq is looking for a lean team of brilliant engineers building powerful infrastructure with an “AI-first” mindset that’s not weighed down by heavy sales or marketing overhead.
“We’re looking for young companies that built the technology, and they want to join a fast-growing company,” Smadari said.
Smadari expects the Revrod acquisition will reduce how quickly users can build and deploy use cases, increase how many incidents an analyst can handle per hour, cut the average time to resolve common incidents such as phishing and minimize the time users spend correcting AI output. Smadari said Torq has a 70% win rate versus Palo Alto Networks and Splunk due to its no-code platform and deep integrations.
“How fast can someone can think about use cases and build something that can result in automation?” Smadari said. “If someone wants to deal with phishing and is building an automation in three days, it will shorten this time frame to a few hours of production.”