Artemis Gets $70M to Build AI Agents for Detection, Response #AI

See Also: AI Security Risks Rise With Agentic Systems

The Felicis-led Series A funding round will enable New York-based Artemis to correlate data across cloud environments, identity systems, networks, endpoints and applications to better understand attacks as they unfold, said co-founder and CEO Shachar Hirshberg. This will overcome the fragmentation inherent in traditional security architectures, where tools fail to provide a cohesive view of attacker behavior.

“We already have the basic product deployed in some of the largest environments in the world, as well as the some of the most sophisticated environments in the world, and customers have already started getting to us inbound,” Hirshberg told ISMG.

Artemis, founded in 2025, employs 30 people and has completed a $15 million seed round led by First Round Capital and Brightmind as well as a $55 million Series A round. The company has been led since its inception by Hirshberg, who previously led product management for AWS’ GuardDuty threat detection service for more than three years (see: AWS Snags Skyhigh’s Gee Rittenhouse to Run Security Business).

Tuning Detections Based on Customer’s Specific Environment

Unlike static detection rules that are applied uniformly across organizations, Artemis dynamically generates and tunes detections based on the specific characteristics of each customer’s environment. Hirshberg underscored that even within a single enterprise, different business units may require distinct detection logic due to variations in infrastructure and operations, which will boost detection accuracy.

“We continuously generate and tune detections that are purpose built for this specific customer, and even for some parts of their environment think like a big enterprise,” Hirshberg said.

The use of natural language interfaces simplifies security operations, which Hirshberg sees as a way to eliminate the need for specialized query languages and manual data exploration. By enabling analysts to ask questions and conduct investigations using plain language, Artemis reduces the technical barrier to entry and accelerates workflows that would otherwise require significant expertise and time, he said.

“It feels like they have a system that just does what they want, and they point in the right direction, and things just happen,” Hirshberg said. “They don’t have to spend three days writing queries manually.”

With Artemis, activities that once took weeks such as correlating alerts, building timelines and conducting investigations can now be completed in minutes. Investigations that historically involved multiple tools, manual correlations and lengthy analysis can be automated and streamlined with Artemis, Hirshberg said.

“Artemis helps companies protect and stop attacks in your environment across their technology stack,” Hirshberg said. “We monitor the logs from across older environments – cloud, identity, network, endpoint, firewall, first-party applications – and connect the dots across the different parts of the stack.”

Defining How Much Autonomy Artemis’s System Has

Artemis’s agentic protection involves using AI agents to autonomously handle the full life cycle of threat detection and response from identifying anomalies to investigating incidents and executing remediation actions. While the vision is highly automated, Hirshberg said it’s designed to operate within a framework that still incorporates human oversight.

“We automatically generate and tune detections continuously based on the operations of your specific company, meaning you really get detections that get how you work and how your assets and entities interact with each other,” Hirshberg said. “Then they have a much higher efficacy rate.”

Artemis aims to address these challenges by using AI agents to handle many of the underlying tasks, he said. These agents can normalize data, generate detections and execute queries automatically, reducing the burden on human analysts. Hirshberg said this helps security professionals focus on higher-value activities such as decision-making and strategy rather than repetitive technical work.

“Every organization is different, so detection that is good for a traditional finance institution working mostly on-prem won’t work in an effective way for an AI-native hyper growth cloud-native startup, because they are in two different places right now,” Hirshberg said.

Despite the push toward automation, Hirshberg said humans are central to the process. Artemis helps companies define how much autonomy the system has, enabling a spectrum from advisory guidance to fully automated action. He noted that customers tend to start with guidance-based workflows and gradually increase automation as trust in the system grows, particularly for low-risk actions.

“We are targeting the largest market of cybersecurity,” Hirshberg said. “So, there is a tremendous opportunity to capture that very high need right now.”