Skip to content
Companies

Incalmo

AI agents that red-team networks without an operator

Raised to date

$8.7M

of $10M offered across 1 Form D filing

Still open
$1.3M
Investors
6
Last filed
Jun 8, 2026

About

Incalmo is a Menlo Park, California cybersecurity company founded in 2025 that builds autonomous cybersecurity systems — AI agents that plan and execute network attacks and defenses without a human operator driving each step. It grew out of CEO Brian Singer's Carnegie Mellon PhD research, published with Anthropic, on an LLM-driven system that autonomously red-teamed multi-host networks and reproduced attacks such as the Equifax breach from high-level strategy alone. The long-term thesis is to pit autonomous attackers and defenders against each other at scale so the defenders improve, in the way self-play trained AlphaGo. The company raised an $8.7M seed round after going through Pear VC's PearX program, and was a finalist at the RSAC 2026 Launch Pad.

Incalmo's agents operate at the level of attack strategy rather than individual commands — the system translates a high-level campaign objective into the concrete host-by-host actions needed to carry it out, which is what lets it red-team a whole multi-host network without a human in the loop.

Funding

  • EquityFiled Jun 8, 2026Form D
    $8.7M/ $10M

Product

Incalmo autonomous cybersecurity system

Autonomous security testing

An LLM-driven system that autonomously red-teams networks: it takes a high-level attack strategy, decomposes it into concrete actions across multiple hosts, executes them, and adapts based on what it finds, with the resulting attack data used to train stronger automated defenders.

  • Autonomous multi-host network red teaming
  • Translates high-level attack strategy into executable system actions
  • LLM-assisted planning with adaptive execution

Built for

Enterprise security teams and organizations defending complex multi-host networks

People on file